Human-Computer Interaction – INTERACT 2021: 18th IFIP TC 13 International Conference, Bari, Italy, August 30 – September 3, 2021, Proceedings, Part II ... Applications, incl. Internet/Web, and HCI) 3030856151, 9783030856151

Citation preview

LNCS 12933

Carmelo Ardito · Rosa Lanzilotti · Alessio Malizia · Helen Petrie · Antonio Piccinno · Giuseppe Desolda · Kori Inkpen (Eds.)

Human-Computer Interaction – INTERACT 2021 18th IFIP TC 13 International Conference Bari, Italy, August 30 – September 3, 2021 Proceedings, Part II

Lecture Notes in Computer Science Founding Editors Gerhard Goos Karlsruhe Institute of Technology, Karlsruhe, Germany Juris Hartmanis Cornell University, Ithaca, NY, USA

Editorial Board Members Elisa Bertino Purdue University, West Lafayette, IN, USA Wen Gao Peking University, Beijing, China Bernhard Steffen TU Dortmund University, Dortmund, Germany Gerhard Woeginger RWTH Aachen, Aachen, Germany Moti Yung Columbia University, New York, NY, USA

12933

More information about this subseries at http://www.springer.com/series/7409

Carmelo Ardito Rosa Lanzilotti Alessio Malizia Helen Petrie Antonio Piccinno Giuseppe Desolda Kori Inkpen (Eds.) •

•

•

•

•

•

Human-Computer Interaction – INTERACT 2021 18th IFIP TC 13 International Conference Bari, Italy, August 30 – September 3, 2021 Proceedings, Part II

123

Editors Carmelo Ardito Department of Electrical and Information Engineering Polytechnic University of Bari Bari, Italy Alessio Malizia Computer Science Department University of Pisa Pisa, Italy University of Hertfordshire Hatfield, United Kingdom Antonio Piccinno Computer Science Department University of Bari Aldo Moro Bari, Italy

Rosa Lanzilotti Computer Science Department University of Bari Aldo Moro Bari, Italy Helen Petrie Department of Computer Science University of York York, UK Giuseppe Desolda Computer Science Department University of Bari Aldo Moro Bari, Italy

Kori Inkpen Microsoft Research Redmond, WA, USA

ISSN 0302-9743 ISSN 1611-3349 (electronic) Lecture Notes in Computer Science ISBN 978-3-030-85615-1 ISBN 978-3-030-85616-8 (eBook) https://doi.org/10.1007/978-3-030-85616-8 LNCS Sublibrary: SL3 – Information Systems and Applications, incl. Internet/Web, and HCI © IFIP International Federation for Information Processing 2021 Chapter “Stress Out: Translating Real-World Stressors into Audio-Visual Stress Cues in VR for Police Training” is licensed under the terms of the Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/). For further details see license information in the chapter. This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Welcome

It is our great pleasure to welcome you to the 18th IFIP TC13 International Conference on Human-Computer Interaction, INTERACT 2021, one of the most important conferences in the area of Human-Computer Interaction at a world-wide level. INTERACT 2021 was held in Bari (Italy) from August 30 – September 3, 2021, in cooperation with ACM and under the patronage of the University of Bari Aldo Moro. This is the second time that INTERACT was held in Italy, after the edition in Rome in September 2005. The Villa Romanazzi Carducci Hotel, which hosted INTERACT 2021, provided the right context for welcoming the participants, thanks to its liberty-period villa immersed in a beautiful park. Due to the COVID-19 pandemic, INTERACT 2021 was held in hybrid mode to allow attendees who could not travel to participate in the conference. INTERACT is held every two years and is well appreciated by the international community, attracting experts with a broad range of backgrounds, coming from all over the world and sharing a common interest in HCI, to make technology effective and useful for all people in their daily life. The theme of INTERACT 2021, “Sense, Feel, Design,” highlighted the new interaction design challenges. Technology is today more and more widespread, pervasive and blended in the world we live in. On one side, devices that sense humans’ activities have the potential to provide an enriched interaction. On the other side, the user experience can be further enhanced by exploiting multisensorial technologies. The traditional human senses of vision and hearing and senses of touch, smell, taste, and emotions can be taken into account when designing for future interactions. The hot topic of this edition was Human-Centered Artificial Intelligence, which implies considering who AI systems are built for and evaluating how well these systems support people’s goals and activities. There was also considerable attention paid to the usable security theme. Not surprisingly, the COVID-19 pandemic and social distancing have also turned the attention of HCI researchers towards the difficulties in performing user-centered design activities and the modified social aspects of interaction. With this, we welcome you all to INTERACT 2021. Several people worked hard to make this conference as pleasurable as possible, and we hope you will truly enjoy it. Paolo Buono Catherine Plaisant

Preface

The 18th IFIP TC13 International Conference on Human-Computer Interaction, INTERACT 2021 (Bari, August 30 – September 3, 2021) attracted a relevant collection of submissions on different topics. Excellent research is the heart of a good conference. Like its predecessors, INTERACT 2021 aimed to foster high-quality research. As a multidisciplinary field, HCI requires interaction and discussion among diverse people with different interests and backgrounds. The beginners and the experienced theoreticians and practitioners, and people from various disciplines and different countries gathered, both in-person and virtually, to learn from each other and contribute to each other’s growth. We were especially honoured to welcome our invited speakers: Marianna Obrist (University College London), Ben Shneiderman (University of Maryland), Luca Viganò (King’s College London), Geraldine Fitzpatrick (TU Wien) and Philippe Palanque (University Toulouse 3 “Paul Sabatier”). Marianna Obrist’s talk focused on the multisensory world people live in and discussed the role touch, taste and smell can play in the future of computing. Ben Shneiderman envisioned a new synthesis of emerging disciplines in which AI-based intelligent algorithms are combined with human-centered design thinking. Luca Viganò used a cybersecurity show and tell approach to illustrate how to use films and other artworks to explain cybersecurity notions. Geraldine Fitzpatrick focused on skills required to use technologies as enablers for good technical design work. Philippe Palanque discussed the cases of system faults due to human errors and presented multiple examples of faults affecting socio-technical systems. A total of 680 submissions, distributed in 2 peer-reviewed tracks, 4 curated tracks, and 3 juried tracks, were received. Of these, the following contributions were accepted: • • • • • • • • •

105 Full Papers (peer-reviewed) 72 Short Papers (peer-reviewed) 36 Posters (juried) 5 Interactive Demos (curated) 9 Industrial Experiences (curated) 3 Panels (curated) 1 Course (curated) 11 Workshops (juried) 13 Doctoral Consortium (juried)

The acceptance rate for contributions received in the peer-reviewed tracks was 29% for full papers and 30% for short papers. In the spirit of inclusiveness of INTERACT, and IFIP in general, a substantial number of promising but borderline full papers, which had not received a direct acceptance decision, were screened for shepherding.

viii

Preface

Interestingly, many of these papers eventually turned out to be excellent quality papers and were included in the final set of full papers. In addition to full papers and short papers, the present proceedings feature’s contributions accepted in the shape of posters, interactive demonstrations, industrial experiences, panels, courses, and descriptions of accepted workshops. Subcommittees managed the reviewing process of the full papers. Each subcommittee had a chair and a set of associated chairs who were in charge of coordinating the reviewing process with the help of expert reviewers. Two new sub-committees were introduced in this edition: “Human-AI Interaction” and “HCI in the Pandemic”. Hereafter we list the sub-committees of INTERACT 2021: • • • • • • • • • • •

Accessibility and assistive technologies Design for business and safety-critical interactive systems Design of interactive entertainment systems HCI education and curriculum HCI in the pandemic Human-AI interaction Information visualization Interactive systems technologies and engineering Methodologies for HCI Social and ubiquitous interaction Understanding users and human behaviour

The final decision on acceptance or rejection of full papers was taken in a Programme Committee meeting held virtually, due to the COVID-19 pandemic, in March 2021. The technical program chairs, the full papers chairs, the subcommittee chairs, and the associate chairs participated in this meeting. The meeting discussed a consistent set of criteria to deal with inevitable differences among many reviewers. The corresponding track chairs and reviewers made the final decisions on other tracks, often after electronic meetings and discussions. We would like to express our strong gratitude to all the people whose passionate and strenuous work ensured the quality of the INTERACT 2021 program: the 12 sub-committees chairs, 88 associated chairs, 34 track chairs, and 543 reviewers; the Keynote & Invited Talks Chair Maria Francesca Costabile; the Posters Chairs Maristella Matera, Kent Norman, Anna Spagnolli; the Interactive Demos Chairs Barbara Rita Barricelli and Nuno Jardim Nunes; the Workshops Chairs Marta Kristín Larusdottir and Davide Spano; the Courses Chairs Nikolaos Avouris and Carmen Santoro; the Panels Chairs Effie Lai-Chong Law and Massimo Zancanaro; the Doctoral Consortium Chairs Daniela Fogli, David Lamas and John Stasko; the Industrial Experiences Chair Danilo Caivano; the Online Experience Chairs Fabrizio Balducci and Miguel Ceriani; the Advisors Fernando Loizides and Marco Winckler; the Student Volunteers Chairs Vita Santa Barletta and Grazia Ragone; the Publicity Chairs Ganesh D. Bhutkar and Veronica Rossano; the Local Organisation Chair Simona Sarti.

Preface

ix

We would like to thank all the authors, who chose INTERACT 2021 as the venue to publish their research and enthusiastically shared their results with the INTERACT community. Last, but not least, we are also grateful to the sponsors for their financial support. Carmelo Ardito Rosa Lanzilotti Alessio Malizia Helen Petrie Antonio Piccinno Giuseppe Desolda Kori Inkpen

IFIP TC13 – http://ifip-tc13.org/

Established in 1989, the Technical Committee on Human–Computer Interaction (IFIP TC 13) of the International Federation for Information Processing (IFIP) is an international committee of 34 member national societies and 10 Working Groups, representing specialists of the various disciplines contributing to the field of human– computer interaction. This includes (among others) human factors, ergonomics, cognitive science, and multiple areas of computer science and design. IFIP TC 13 aims to develop the science, technology and societal aspects of human– computer interaction (HCI) by • encouraging empirical, applied and theoretical research • promoting the use of knowledge and methods from both human sciences and computer sciences in design, development, evaluation and exploitation of computing systems • promoting the production of new knowledge in the area of interactive computing systems engineering • promoting better understanding of the relation between formal design methods and system usability, user experience, accessibility and acceptability • developing guidelines, models and methods by which designers may provide better human-oriented computing systems • and, cooperating with other groups, inside and outside IFIP, to promote userorientation and humanization in system design. Thus, TC 13 seeks to improve interactions between people and computing systems, to encourage the growth of HCI research and its practice in industry and to disseminate these benefits worldwide. The main orientation is to place the users at the center of the development process. Areas of study include: • the problems people face when interacting with computing devices; • the impact of technology deployment on people in individual and organizational contexts; • the determinants of utility, usability, acceptability, accessibility, privacy, and user experience ...; • the appropriate allocation of tasks between computing systems and users especially in the case of automation; • engineering user interfaces, interactions and interactive computing systems; • modelling the user, their tasks and the interactive system to aid better system design; and harmonizing the computing system to user characteristics and needs. While the scope is thus set wide, with a tendency toward general principles rather than particular systems, it is recognized that progress will only be achieved through

xii

IFIP TC13 – http://ifip-tc13.org/

both general studies to advance theoretical understandings and specific studies on practical issues (e.g., interface design standards, software system resilience, documentation, training material, appropriateness of alternative interaction technologies, guidelines, integrating computing systems to match user needs and organizational practices, etc.). In 2015, TC13 approved the creation of a Steering Committee (SC) for the INTERACT conference series. The SC is now in place, chaired by Anirudha Joshi and is responsible for: • promoting and maintaining the INTERACT conference as the premiere venue for researchers and practitioners interested in the topics of the conference (this requires a refinement of the topics above); • ensuring the highest quality for the contents of the event; • setting up the bidding process to handle the future INTERACT conferences (decision is made at TC 13 level); • providing advice to the current and future chairs and organizers of the INTERACT conference; • providing data, tools, and documents about previous conferences to the future conference organizers; • selecting the reviewing system to be used throughout the conference (as this affects the entire set of reviewers, authors and committee members); • resolving general issues involved with the INTERACT conference; • capitalizing on history (good and bad practices). In 1999, TC 13 initiated a special IFIP Award, the Brian Shackel Award, for the most outstanding contribution in the form of a refereed paper submitted to and delivered at each INTERACT. The award draws attention to the need for a comprehensive human-centered approach in the design and use of information technology in which the human and social implications have been taken into account. In 2007, IFIP TC 13 launched an Accessibility Award to recognize an outstanding contribution in HCI with international impact dedicated to the field of accessibility for disabled users. In 2013, IFIP TC 13 launched the Interaction Design for International Development (IDID) Award that recognizes the most outstanding contribution to the application of interactive systems for social and economic development of people in developing countries. Since the process to decide the award takes place after papers are sent to the publisher for publication, the awards are not identified in the proceedings. Since 2019 a special agreement has been made with the International Journal of Behaviour & Information Technology (published by Taylor & Francis) with Panos Markopoulos as editor in chief. In this agreement, authors of BIT whose papers are within the field of HCI are offered the opportunity to present their work at the INTERACT conference. Reciprocally, a selection of papers submitted and accepted for presentation at INTERACT are offered the opportunity to extend their contribution to be published in BIT. IFIP TC 13 also recognizes pioneers in the area of HCI. An IFIP TC 13 pioneer is one who, through active participation in IFIP Technical Committees or related IFIP groups, has made outstanding contributions to the educational, theoretical, technical, commercial, or professional aspects of analysis, design, construction, evaluation, and

IFIP TC13 – http://ifip-tc13.org/

xiii

use of interactive systems. IFIP TC 13 pioneers are appointed annually and awards are handed over at the INTERACT conference. IFIP TC 13 stimulates working events and activities through its Working Groups (WGs). Working Groups consist of HCI experts from multiple countries, who seek to expand knowledge and find solutions to HCI issues and concerns within a specific domain. The list of Working Groups and their domains is given below. WG13.1 (Education in HCI and HCI Curricula) aims to improve HCI education at all levels of higher education, coordinate and unite efforts to develop HCI curricula and promote HCI teaching. WG13.2 (Methodology for User-Centred System Design) aims to foster research, dissemination of information and good practice in the methodical application of HCI to software engineering. WG13.3 (HCI, Disability and Aging) aims to make HCI designers aware of the needs of people with disabilities and encourage development of information systems and tools permitting adaptation of interfaces to specific users. WG13.4 (also WG2.7) (User Interface Engineering) investigates the nature, concepts and construction of user interfaces for software systems, using a framework for reasoning about interactive systems and an engineering model for developing UIs. WG 13.5 (Resilience, Reliability, Safety and Human Error in System Development) seeks a framework for studying human factors relating to systems failure, develops leading edge techniques in hazard analysis and safety engineering of computer-based systems, and guides international accreditation activities for safety-critical systems. WG13.6 (Human-Work Interaction Design) aims at establishing relationships between extensive empirical work-domain studies and HCI design. It will promote the use of knowledge, concepts, methods and techniques that enable user studies to procure a better apprehension of the complex interplay between individual, social and organizational contexts and thereby a better understanding of how and why people work in the ways that they do. WG13.7 (Human–Computer Interaction and Visualization) aims to establish a study and research program that will combine both scientific work and practical applications in the fields of human–computer interaction and visualization. It will integrate several additional aspects of further research areas, such as scientific visualization, data mining, information design, computer graphics, cognition sciences, perception theory, or psychology, into this approach. WG13.8 (Interaction Design and International Development) is currently working to reformulate their aims and scope. WG13.9 (Interaction Design and Children) aims to support practitioners, regulators and researchers to develop the study of interaction design and children across international contexts. WG13.10 (Human-Centred Technology for Sustainability) aims to promote research, design, development, evaluation, and deployment of human-centered technology to encourage sustainable use of resources in various domains. New Working Groups are formed as areas of significance in HCI arise. Further information is available at the IFIP TC13 website: http://ifip-tc13.org/.

IFIP TC13 Members

Officers Chairperson

Vice-chair for Working Groups

Philippe Palanque, France

Simone D. J. Barbosa, Brazil

Vice-chair for Awards

Vice-chair for Development and Equity

Paula Kotze, South Africa

Julio Abascal, Spain

Vice-chair for Communications

Treasurer

Helen Petrie, UK

Virpi Roto, Finland

Vice-chair for Growth and Reach out INTERACT Steering Committee Chair

Secretary

Jan Gulliksen, Sweden

INTERACT Steering Committee Chair

Marco Winckler, France

Anirudha Joshi, India

Country Representatives Australia Henry B. L. Duh Australian Computer Society Austria Geraldine Fitzpatrick Austrian Computer Society Belgium Bruno Dumas IMEC – Interuniversity Micro-Electronics Center Brazil Lara S. G. Piccolo Brazilian Computer Society (SBC) Bulgaria Stoyan Georgiev Dentchev Bulgarian Academy of Sciences

Croatia Andrina Granic Croatian Information Technology Association (CITA) Cyprus Panayiotis Zaphiris Cyprus Computer Society Czech Republic Zdeněk Míkovec Czech Society for Cybernetics and Informatics Finland Virpi Roto Finnish Information Processing Association

xvi

IFIP TC13 Members

France Philippe Palanque and Marco Winckler Société informatique de France (SIF)

Singapore Shengdong Zhao Singapore Computer Society

Germany Tom Gross Gesellschaft fur Informatik e.V.

Slovakia Wanda Benešová The Slovak Society Science

Ireland Liam J. Bannon Irish Computer Society Italy Fabio Paternò Italian Computer Society Japan Yoshifumi Kitamura Information Processing Society of Japan Netherlands Regina Bernhaupt Nederlands Genootschap voor Informatica New Zealand Mark Apperley New Zealand Computer Society Norway Frode Eika Sandnes Norwegian Computer Society Poland Marcin Sikorski Poland Academy of Sciences Portugal Pedro Campos Associacão Portuguesa para o Desenvolvimento da Sociedade da Informação (APDSI) Serbia Aleksandar Jevremovic Informatics Association of Serbia

Slovenia Matjaž Debevc The Slovenian INFORMATIKA

for

Computer

Computer

Society

Sri Lanka Thilina Halloluwa The Computer Society of Sri Lanka South Africa Janet L. Wesson & Paula Kotze The Computer Society of South Africa Sweden Jan Gulliksen Swedish Interdisciplinary Society for Human-Computer Interaction Swedish Computer Society Switzerland Denis Lalanne Swiss Federation for Information Processing Tunisia Mona Laroussi Ecole Supérieure des Communications de Tunis (SUP’COM) United Kingdom José Abdelnour Nocera British Computer Society (BCS) United Arab Emirates Ahmed Seffah UAE Computer Society

IFIP TC13 Members

xvii

ACM

CLEI

Gerrit van der Veer Association for Computing Machinery

Jaime Sánchez Centro Latinoamericano de Estudios en Informatica

Expert Members Julio Abascal, Spain Carmelo Ardito, Italy Nikolaos Avouris, Greece Kaveh Bazargan, Iran Ivan Burmistrov, Russia Torkil Torkil Clemmensen, Denmark Peter Forbrig, Germany Dorian Gorgan, Romania

Anirudha Joshi, India David Lamas, Estonia Marta Kristin Larusdottir, Iceland Zhengjie Liu, China Fernando Loizides, UK/Cyprus Ochieng Daniel “Dan” Orwa, Kenya Eunice Sari, Australia/Indonesia

Working Group Chairpersons WG 13.1 (Education in HCI and HCI Curricula)

WG13.6 (Human-Work Interaction Design)

Konrad Baumann, Austria

Barbara Rita Barricelli, Italy

WG 13.2 (Methodologies for User-Centered System Design)

WG13.7 (HCI and Visualization) Peter Dannenmann, Germany

Regina Bernhaupt, Netherlands WG 13.3 (HCI, Disability and Aging)

WG 13.8 (Interaction Design and International Development)

Helen Petrie, UK

José Adbelnour Nocera, UK

WG 13.4/2.7 (User Interface Engineering)

WG 13.9 (Interaction Design and Children)

José Creissac Campos, Portugal

Janet Read, UK

WG 13.5 (Human Error, Resilience, Reliability, Safety and System Development)

WG 13.10 (Human-Centred Technology for Sustainability)

Chris Johnson, UK

Masood Masoodian, Finland

Conference Organizing Committee

General Conference Co-chairs

Courses Co-chairs

Paolo Buono, Italy Catherine Plaisant, USA and France

Carmen Santoro, Italy Nikolaos Avouris, Greece

Advisors

Industrial Experiences Chair

Fernando Loizides, UK Marco Winckler, France

Danilo Caivano, Italy Workshops Co-chairs

Technical Program Co-chairs Carmelo Ardito, Italy Rosa Lanzilotti, Italy Alessio Malizia, UK and Italy

Marta Kristín Larusdottir, Iceland Davide Spano, Italy Doctoral Consortium Co-chairs

Maria Francesca Costabile, Italy

Daniela Fogli, Italy David Lamas, Estonia John Stasko, USA

Full Papers Co-chairs

Online Experience Co-chairs

Helen Petrie, UK Antonio Piccinno, Italy

Fabrizio Balducci, Italy Miguel Ceriani, Italy

Short Papers Co-chairs

Student Volunteers Co-chairs

Giuseppe Desolda, Italy Kori Inkpen, USA

Vita Santa Barletta, Italy Grazia Ragone, UK

Posters Co-chairs

Publicity Co-chairs

Maristella Matera, Italy Kent Norman, USA Anna Spagnolli, Italy

Ganesh D. Bhutkar, India Veronica Rossano, Italy

Keynote and Invited Talks Chair

Local Organisation Chair Interactive Demos Co-chairs Barbara Rita Barricelli, Italy Nuno Jardim Nunes, Portugal Panels Co-chairs Effie Lai-Chong Law, UK Massimo Zancanaro, Italy

Simona Sarti, Consulta Umbria, Italy

xx

Conference Organizing Committee

Programme Committee Sub-committee Chairs Nikolaos Avouris, Greece Regina Bernhaupt, Netherlands Carla Dal Sasso Freitas, Brazil Jan Gulliksen, Sweden Paula Kotzé, South Africa Effie Lai-Chong Law, UK

Philippe Palanque, France Fabio Paternò, Italy Thomas Pederson, Sweden Albrecht Schmidt, Germany Frank Steinicke, Germany Gerhard Weber, Germany

Associated Chairs José Abdelnour Nocera, UK Raian Ali, Qatar Florian Alt, Germany Katrina Attwood, UK Simone Barbosa, Brazil Cristian Bogdan, Sweden Paolo Bottoni, Italy Judy Bowen, New Zealand Daniel Buschek, Germany Pedro Campos, Portugal José Creissac Campos, Portugal Luca Chittaro, Italy Sandy Claes, Belgium Christopher Clarke, UK Torkil Clemmensen, Denmark Vanessa Cobus, Germany Ashley Colley, Finland Aurora Constantin, UK Lynne Coventry, UK Yngve Dahl, Norway Maria De Marsico, Italy Luigi De Russis, Italy Paloma Diaz, Spain Monica Divitini, Norway Mateusz Dolata, Switzerland Bruno Dumas, Belgium Sophie Dupuy-Chessa, France Dan Fitton, UK Peter Forbrig, Germany Sandnes Frode Eika, Norway Vivian Genaro Motti, USA Rosella Gennari, Italy

Jens Gerken, Germany Mareike Glöss, Sweden Dorian Gorgan, Romania Tom Gross, Germany Uwe Gruenefeld, Germany Julie Haney, USA Ebba Þóra Hvannberg, Iceland Netta Iivari, Finland Nanna Inie, Denmark Anna Sigríður Islind, Iceland Anirudha Joshi, India Bridget Kane, Sweden Anne Marie Kanstrup, Denmark Mohamed Khamis, UK Kibum Kim, Korea Marion Koelle, Germany Kati Kuusinen, Denmark Matthias Laschke, Germany Fernando Loizides, UK Andrés Lucero, Finland Jo Lumsden, UK Charlotte Magnusson, Sweden Andrea Marrella, Italy Célia Martinie, France Timothy Merritt, Denmark Zdeněk Míkovec, Czech Republic Luciana Nedel, Brazil Laurence Nigay, France Valentina Nisi, Portugal Raquel O. Prates, Brazil Rakesh Patibanda, Australia Simon Perrault, Singapore

Conference Organizing Committee

Lara Piccolo, UK Aparecido Fabiano Pinatti de Carvalho, Germany Janet Read, UK Karen Renaud, UK Antonio Rizzo, Italy Sayan Sarcar, Japan Valentin Schwind, Germany Gavin Sim, UK Fotios Spyridonis, UK Jan Stage, Denmark Simone Stumpf, UK Luis Teixeira, Portugal

Jakob Tholander, Sweden Daniela Trevisan, Brazil Stefano Valtolina, Italy Jan Van den Bergh, Belgium Nervo Verdezoto, UK Chi Vi, UK Giuliana Vitiello, Italy Sarah Völkel, Germany Marco Winckler, France Dhaval Vyas, Australia Janet Wesson, South Africa Paweł W. Woźniak, Netherlands

Reviewers Bruno A. Chagas, Brazil Yasmeen Abdrabou, Germany Maher Abujelala, USA Jiban Adhikary, USA Kashif Ahmad, Qatar Muneeb Ahmad, UK Naveed Ahmed, United Arab Emirates Aino Ahtinen, Finland Wolfgang Aigner, Austria Deepak Akkil, Finland Aftab Alam, Republic of Korea Soraia Meneses Alarcão, Portugal Pedro Albuquerque Santos, Portugal Günter Alce, Sweden Iñigo Aldalur, Spain Alaa Alkhafaji, Iraq Aishat Aloba, USA Yosuef Alotaibi, UK Taghreed Alshehri, UK Ragaad Al-Tarawneh, USA Alejandro Alvarez-Marin, Chile Lucas Anastasiou, UK Ulf Andersson, Sweden Joseph Aneke, Italy Mark Apperley, New Zealand Renan Aranha, Brazil Pierre-Emmanuel Arduin, France Stephanie Arevalo Arboleda, Germany Jan Argasiński, Poland

Patricia Arias-Cabarcos, Germany Alexander Arntz, Germany Jonas Auda, Germany Andreas Auinger, Austria Iuliia Avgustis, Finland Cédric Bach, France Miroslav Bachinski, Germany Victor Bacu, Romania Jan Balata, Czech Republic Teresa Baldassarre, Italy Fabrizio Balducci, Italy Vijayanand Banahatti, India Karolina Baras, Portugal Simone Barbosa, Brazil Vita Santa Barletta, Italy Silvio Barra, Italy Barbara Rita Barricelli, Italy Ralph Barthel, UK Thomas Baudel, France Christine Bauer, Netherlands Fatma Ben Mesmia, Canada Marit Bentvelzen, Netherlands François Bérard, France Melanie Berger, Netherlands Gerd Berget, Norway Sergi Bermúdez i Badia, Portugal Dario Bertero, UK Guilherme Bertolaccini, Brazil Lonni Besançon, Australia

xxi

xxii

Conference Organizing Committee

Laura-Bianca Bilius, Romania Kerstin Blumenstein, Austria Andreas Bollin, Austria Judith Borghouts, UK Nis Bornoe, Denmark Gabriela Bosetti, UK Hollie Bostock, Portugal Paolo Bottoni, Italy Magdalena Boucher, Austria Amina Bouraoui, Tunisia Elodie Bouzekri, France Judy Bowen, New Zealand Efe Bozkir, Germany Danielle Bragg, USA Diogo Branco, Portugal Dawn Branley-Bell, UK Stephen Brewster, UK Giada Brianza, UK Barry Brown, Sweden Nick Bryan-Kinns, UK Andreas Bucher, Switzerland Elizabeth Buie, UK Alexandru Bundea, Germany Paolo Buono, Italy Michael Burch, Switzerland Matthew Butler, Australia Fabio Buttussi, Italy Andreas Butz, Germany Maria Claudia Buzzi, Italy Marina Buzzi, Italy Zoya Bylinskii, USA Diogo Cabral, Portugal Åsa Cajander, Sweden Francisco Maria Calisto, Portugal Hector Caltenco, Sweden José Creissac Campos, Portugal Heloisa Candello, Brazil Alberto Cannavò, Italy Bruno Cardoso, Belgium Jorge Cardoso, Portugal Géry Casiez, France Fabio Cassano, Italy Brendan Cassidy, UK Alejandro Catala, Spain

Miguel Ceriani, UK Daniel Cermak-Sassenrath, Denmark Vanessa Cesário, Portugal Fred Charles, UK Debaleena Chattopadhyay, USA Alex Chen, Singapore Thomas Chen, USA Yuan Chen, Canada Chola Chhetri, USA Katherine Chiluiza, Ecuador Nick Chozos, UK Michael Chromik, Germany Christopher Clarke, UK Bárbara Cleto, Portugal Antonio Coelho, Portugal Ashley Colley, Finland Nelly Condori-Fernandez, Spain Marios Constantinides, UK Cléber Corrêa, Brazil Vinicius Costa de Souza, Brazil Joëlle Coutaz, France Céline Coutrix, France Chris Creed, UK Carlos Cunha, Portugal Kamila Rios da Hora Rodrigues, Brazil Damon Daylamani-Zad, UK Sergio de Cesare, UK Marco de Gemmis, Italy Teis De Greve, Belgium Victor Adriel de Jesus Oliveira, Austria Helmut Degen, USA Donald Degraen, Germany William Delamare, France Giuseppe Desolda, Italy Henrik Detjen, Germany Marianna Di Gregorio, Italy Ines Di Loreto, France Daniel Diethei, Germany Tilman Dingler, Australia Anke Dittmar, Germany Monica Divitini, Norway Janki Dodiya, Germany Julia Dominiak, Poland Ralf Dörner, Germany

Conference Organizing Committee

Julie Doyle, Ireland Philip Doyle, Ireland Fiona Draxler, Germany Emanuel Felipe Duarte, Brazil Rui Duarte, Portugal Bruno Dumas, Belgium Mark Dunlop, UK Sophie Dupuy-Chessa, France Jason Dykes, UK Chloe Eghtebas, Germany Kevin El Haddad, Belgium Don Samitha Elvitigala, New Zealand Augusto Esteves, Portugal Siri Fagernes, Norway Katherine Fennedy, Singapore Marta Ferreira, Portugal Francesco Ferrise, Italy Lauren Stacey Ferro, Italy Christos Fidas, Greece Daniel Finnegan, UK Daniela Fogli, Italy Manuel J. Fonseca, Portugal Peter Forbrig, Germany Rita Francese, Italy André Freire, Brazil Karin Fröhlich, Finland Susanne Furman, USA Henrique Galvan Debarba, Denmark Sandra Gama, Portugal Dilrukshi Gamage, Japan Jérémie Garcia, France Jose Garcia Estrada, Norway David Geerts, Belgium Denise Y. Geiskkovitch, Canada Stefan Geisler, Germany Mirko Gelsomini, Italy Çağlar Genç, Finland Rosella Gennari, Italy Nina Gerber, Germany Moojan Ghafurian, Canada Maliheh Ghajargar, Sweden Sabiha Ghellal, Germany Debjyoti Ghosh, Germany Michail Giannakos, Norway

xxiii

Terje Gjøsæter, Norway Marc Gonzalez Capdevila, Brazil Julien Gori, Finland Laurent Grisoni, France Tor-Morten Gronli, Norway Sebastian Günther, Germany Li Guo, UK Srishti Gupta, USA Francisco Gutiérrez, Belgium José Eder Guzman Mendoza, Mexico Jonna Häkkilä, Finland Lilit Hakobyan, UK Thilina Halloluwa, Sri Lanka Perttu Hämäläinen, Finland Lane Harrison, USA Michael Harrison, UK Hanna Hasselqvist, Sweden Tomi Heimonen, USA Florian Heinrich, Germany Florian Heller, Belgium Karey Helms, Sweden Nathalie Henry Riche, USA Diana Hernandez-Bocanegra, Germany Danula Hettiachchi, Australia Wilko Heuten, Germany Annika Hinze, New Zealand Linda Hirsch, Germany Sarah Hodge, UK Sven Hoffmann, Germany Catherine Holloway, UK Leona Holloway, Australia Lars Erik Holmquist, UK Anca-Simona Horvath, Denmark Simo Hosio, Finland Sebastian Hubenschmid, Germany Helena Vallo Hult, Sweden Shah Rukh Humayoun, USA Ebba Þóra Hvannberg, Iceland Alon Ilsar, Australia Md Athar Imtiaz, New Zealand Oana Inel, Netherlands Francisco Iniesto, UK Andri Ioannou, Cyprus Chyng-Yang Jang, USA

xxiv

Conference Organizing Committee

Gokul Jayakrishnan, India Stine Johansen, Denmark Tero Jokela, Finland Rui José, Portugal Anirudha Joshi, India Manjiri Joshi, India Jana Jost, Germany Patrick Jost, Norway Annika Kaltenhauser, Germany Jin Kang, Canada Younah Kang, Republic of Korea Jari Kangas, Finland Petko Karadechev, Denmark Armağan Karahanoğlu, Netherlands Sukran Karaosmanoglu, Germany Alexander Kempton, Norway Rajiv Khadka, USA Jayden Khakurel, Finland Pramod Khambete, India Neeta Khanuja, Portugal Young-Ho Kim, USA Reuben Kirkham, Australia Ilan Kirsh, Israel Maria Kjærup, Denmark Kevin Koban, Austria Frederik Kobbelgaard, Denmark Martin Kocur, Germany Marius Koller, Germany Christophe Kolski, France Takanori Komatsu, Japan Jemma König, New Zealand Monika Kornacka, Poland Thomas Kosch, Germany Panayiotis Koutsabasis, Greece Lucie Kruse, Germany Przemysław Kucharski, Poland Johannes Kunkel, Germany Bineeth Kuriakose, Norway Anelia Kurteva, Austria Marc Kurz, Austria Florian Lang, Germany Rosa Lanzilotti, Italy Lars Bo Larsen, Denmark Marta Larusdottir, Iceland

Effie Law, UK Luis Leiva, Luxembourg Barbara Leporini, Italy Pascal Lessel, Germany Hongyu Li, USA Yuan Liang, USA Yu-Tzu Lin, Denmark Markus Löchtefeld, Denmark Angela Locoro, Italy Benedikt Loepp, Germany Domenico Lofù, Italy Fernando Loizides, UK Arminda Lopes, Portugal Feiyu Lu, USA Jo Lumsden, UK Anders Lundström, Sweden Kris Luyten, Belgium Granit Luzhnica, Austria Marc Macé, France Anderson Maciel, Brazil Cristiano Maciel, Brazil Miroslav Macík, Czech Republic Scott MacKenzie, Canada Hanuma Teja Maddali, USA Rui Madeira, Portugal Alexander Maedche, Germany Charlotte Magnusson, Sweden Jyotirmaya Mahapatra, India Vanessa Maike, USA Maitreyee Maitreyee, Sweden Ville Mäkelä, Germany Sylvain Malacria, France Sugandh Malhotra, India Ivo Malý, Czech Republic Marco Manca, Italy Muhanad Manshad, USA Isabel Manssour, Brazil Panos Markopoulos, Netherlands Karola Marky, Germany Andrea Marrella, Italy Andreas Martin, Switzerland Célia Martinie, France Nuno Martins, Portugal Maristella Matera, Italy

Conference Organizing Committee

Florian Mathis, UK Andrii Matviienko, Germany Peter Mayer, Germany Sven Mayer, Germany Mark McGill, UK Donald McMillan, Sweden Lukas Mecke, Germany Elisa Mekler, Finland Alessandra Melonio, Italy Eleonora Mencarini, Italy Maria Menendez Blanco, Italy Aline Menin, France Arjun Menon, Sweden Nazmus Sakib Miazi, USA Zdeněk Míkovec, Czech Republic Tim Mittermeier, Germany Emmanuel Mkpojiogu, Nigeria Jonas Moll, Sweden Alberto Monge Roffarello, Italy Troels Mønsted, Denmark Diego Morra, Italy Jaime Munoz Arteaga, Mexico Sachith Muthukumarana, New Zealand Vasiliki Mylonopoulou, Sweden Frank Nack, Netherlands Mohammad Naiseh, UK Vania Neris, Brazil Robin Neuhaus, Germany Thao Ngo, Germany Binh Vinh Duc Nguyen, Belgium Vickie Nguyen, USA James Nicholson, UK Peter Axel Nielsen, Denmark Jasmin Niess, Germany Evangelos Niforatos, Netherlands Kent Norman, USA Fatima Nunes, Brazil Carli Ochs, Switzerland Joseph O’Hagan, UK Takashi Ohta, Japan Jonas Oppenlaender, Finland Michael Ortega, France Changkun Ou, Germany Yun Suen Pai, New Zealand

Dominika Palivcová, Czech Republic Viktoria Pammer-Schindler, Austria Eleftherios Papachristos, Denmark Sofia Papavlasopoulou, Norway Leonado Parra, Colombia Max Pascher, Germany Ankit Patel, Portugal Fabio Paternò, Italy Maria Angela Pellegrino, Italy Anthony Perritano, USA Johanna Persson, Sweden Ken Pfeuffer, Germany Bastian Pfleging, Netherlands Vung Pham, USA Jayesh Pillai, India Catherine Plaisant, USA Henning Pohl, Denmark Margit Pohl, Austria Alessandro Pollini, Italy Dorin-Mircea Popovici, Romania Thiago Porcino, Brazil Dominic Potts, UK Sarah Prange, Germany Marco Procaccini, Italy Arnaud Prouzeau, France Parinya Punpongsanon, Japan Sónia Rafael, Portugal Jessica Rahman, Australia Mikko Rajanen, Finland Nimmi Rangaswamy, India Alberto Raposo, Brazil George Raptis, Greece Hanae Rateau, Canada Sebastian Rauh, Germany Hirak Ray, USA Traian Rebedea, Romania Yosra Rekik, France Elizabeth Rendon-Velez, Colombia Malte Ressin, UK Tera Reynolds, USA Miguel Ribeiro, Portugal Maria Rigou, Greece Sirpa Riihiaho, Finland Michele Risi, Italy

xxv

xxvi

Conference Organizing Committee

Radiah Rivu, Germany Mehdi Rizvi, Italy Judy Robertson, UK Michael Rohs, Germany Marco Romano, Italy Anton Rosén, Sweden Veronica Rossano, Italy Virpi Roto, Finland Debjani Roy, India Matthew Rueben, USA Vit Rusnak, Czech Republic Philippa Ryan, UK Thomas Ryberg, Denmark Rufat Rzayev, Germany Parisa Saadati, UK Adrian Sabou, Romania Ofir Sadka, Israel Juan Pablo Saenz, Italy Marco Saltarella, Italy Sanjit Samaddar, UK Ivan Sanchez Milara, Finland Frode Eika Sandnes, Norway Leonardo Sandoval, UK Carmen Santoro, Italy Pratiti Sarkar, India Guilherme Schardong, Brazil Christina Schmidbauer, Austria Eike Schneiders, Denmark Maximilian Schrapel, Germany Sabrina Scuri, Portugal Korok Sengupta, Germany Marta Serafini, Italy Marcos Serrano, France Kshitij Sharma, Norway Sumita Sharma, Finland Akihisa Shitara, Japan Mark Shovman, Israel Ludwig Sidenmark, UK Carlos Silva, Portugal Tiago Silva da Silva, Brazil Milene Silveira, Brazil James Simpson, Australia Ashwin Singh, India Laurianne Sitbon, Australia

Mikael B. Skov, Denmark Pavel Slavik, Czech Republic Aidan Slingsby, UK K. Tara Smith, UK Ellis Solaiman, UK Andreas Sonderegger, Switzerland Erik Sonnleitner, Austria Keyur Sorathia, India Emanuel Sousa, Portugal Sonia Sousa, Estonia Anna Spagnolli, Italy Davide Spallazzo, Italy Lucio Davide Spano, Italy Katta Spiel, Austria Priyanka Srivastava, India Katarzyna Stawarz, UK Teodor Stefanut, Romania Mari-Klara Stein, Denmark Jonathan Strahl, Finland Tim Stratmann, Germany Christian Sturm, Germany Michael Svangren, Denmark Aurélien Tabard, France Benjamin Tag, Australia Federico Tajariol, France Aishwari Talhan, Republic of Korea Maurizio Teli, Denmark Subrata Tikadar, India Helena Tobiasson, Sweden Guy Toko, South Africa Brianna Tomlinson, USA Olof Torgersson, Sweden Genoveffa Tortora, Italy Zachary O. Toups, USA Scott Trent, Japan Philippe Truillet, France Tommaso Turchi, UK Fanny Vainionpää, Finland Stefano Valtolina, Italy Niels van Berkel, Denmark Jan Van den Bergh, Belgium Joey van der Bie, Netherlands Bram van Deurzen, Belgium Domenique van Gennip, Australia

Conference Organizing Committee

Paul van Schauk, UK Koen van Turnhout, Netherlands Jean Vanderdonckt, Belgium Eduardo Veas, Austria Katia Vega, USA Kellie Vella, Australia Leena Ventä-Olkkonen, Finland Nadine Vigouroux, France Gabriela Villalobos-Zúñiga, Switzerland Aku Visuri, Finland Giuliana Vitiello, Italy Pierpaolo Vittorini, Italy Julius von Willich, Germany Steven Vos, Netherlands Nadine Wagener, Germany Lun Wang, Italy Ruijie Wang, Italy Gerhard Weber, Germany Thomas Weber, Germany Rina Wehbe, Canada Florian Weidner, Germany Alexandra Weilenmann, Sweden Sebastian Weiß, Germany

Yannick Weiss, Germany Robin Welsch, Germany Janet Wesson, South Africa Benjamin Weyers, Germany Stephanie Wilson, UK Marco Winckler, France Philipp Wintersberger, Austria Katrin Wolf, Germany Kim Wölfel, Germany Julia Woodward, USA Matthias Wunsch, Austria Haijun Xia, USA Asim Evren Yantac, Turkey Enes Yigitbas, Germany Yongjae Yoo, Canada Johannes Zagermann, Germany Massimo Zancanaro, Italy André Zenner, Germany Jingjie Zheng, Canada Suwen Zhu, USA Ying Zhu, USA Jürgen Ziegler, Germany

xxvii

xxviii

Conference Organizing Committee

Partners and Sponsors Partners

International Federation for Information Processing

In-cooperation with ACM

In-cooperation with SIGCHI

Conference Organizing Committee

Sponsors

xxix

Contents – Part II

COVID-19 and HCI Addressing the Challenges of COVID-19 Social Distancing Through Passive Wi-Fi and Ubiquitous Analytics: A Real World Deployment. . . . . . . Miguel Ribeiro, Nuno Nunes, Marta Ferreira, João Nogueira, Johannes Schöning, and Valentina Nisi Hanging Out Online: Social Life During the Pandemic . . . . . . . . . . . . . . . . Ashwin Singh and Grace Eden

3

25

Investigating Italian Citizens’ Attitudes Towards Immuni, the Italian Contact Tracing App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cristina Bosco and Martina Cvajner

34

Social Companion Robots to Reduce Isolation: A Perception Change Due to COVID-19. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Moojan Ghafurian, Colin Ellard, and Kerstin Dautenhahn

43

Crowdsourcing Methods in HCI BubbleVideo: Supporting Small Group Interactions in Online Conferences. . . Bill Rogers, Mark Apperley, and Masood Masoodian Comparing Performance Models for Bivariate Pointing Through a Crowdsourced Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shota Yamanaka Older Adults’ Motivation and Engagement with Diverse Crowdsourcing Citizen Science Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kinga Skorupska, Anna Jaskulska, Rafał Masłyk, Julia Paluch, Radosław Nielek, and Wiesław Kopeć Quality Assessment of Crowdwork via Eye Gaze: Towards Adaptive Personalized Crowdsourcing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Md. Rabiul Islam, Shun Nawa, Andrew Vargo, Motoi Iwata, Masaki Matsubara, Atsuyuki Morishima, and Koichi Kise

67

76

93

104

Design for Automotive Interfaces Designing for a Convenient In-Car Passenger Experience: A Repertory Grid Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Melanie Berger, Bastian Pfleging, and Regina Bernhaupt

117

xxxii

Contents – Part II

Exploring Application Opportunities for Smart Vehicles in the Continuous Interaction Space Inside and Outside the Vehicle . . . . . . . . . . . . . . . . . . . . Laura-Bianca Bilius, Radu-Daniel Vatavu, and Nicolai Marquardt

140

Smart Vehicle Proxemics: A Conceptual Framework Operationalizing Proxemics in the Context of Outside-the-Vehicle Interactions . . . . . . . . . . . . Laura-Bianca Bilius, Radu-Daniel Vatavu, and Nicolai Marquardt

150

Design Methods Advanced Kidney Disease Patient Portal: Implementation and Evaluation with Haemodialysis Patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramsay Meiklem, Karen Stevenson, Sabine Richarz, David B. Kingsmore, Matt-Mouley Bouamrane, Mark Dunlop, and Peter Thomson

175

Digital Work Environment Rounds – Systematic Inspections of Usability Supported by the Legislation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jan Gulliksen

197

Facilitating User Involvement in a Large IT Project: A Comparison of Facilitators’ Perspectives on Process, Role and Personal Practice. . . . . . . . Øivind Klungseth Zahlsen, Dag Svanæs, and Yngve Dahl

219

Focus, Structure, Reflection! Integrating User-Centred Design and Design Sprint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Virpi Roto, Marta Larusdottir, Andrés Lucero, Jan Stage, and Ilja Šmorgun How HCI Interprets Service Design: A Systematic Literature Review . . . . . . Christine Ee Ling Yap, Jung-Joo Lee, and Virpi Roto Sniff Before You Act: Exploration of Scent-Feature Associations for Designing Future Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Giada Brianza, Patricia Cornelio, Emanuela Maggioni, and Marianna Obrist Tales from the Materialverse: Comic-Based Narratives and Character Cut-Outs for Co-Design Fiction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eleni Economidou, Susanna Vogel, Nathalia Campreguer França, Bernhard Maurer, and Manfred Tscheligi Understanding the Role of Physical and Digital Techniques in the Initial Design Processes of Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emrecan Gulay and Andrés Lucero Understanding Users Through Three Types of Personas . . . . . . . . . . . . . . . . Lene Nielsen, Marta Larusdottir, and Lars Bo Larsen

239

259

281

302

312 330

Contents – Part II

xxxiii

Designing for Smart Devices and IoT Exploring Perceptions of a Localized Content-Sharing System Using Users-as-Beacons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nazmus Sakib Miazi, Heather Lipford, and Mohamed Shehab SENSATION: An Authoring Tool to Support Event–State Paradigm in End-User Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Giuseppe Desolda, Francesco Greco, Francisco Guarnieri, Nicole Mariz, and Massimo Zancanaro The Controversy of Responsibility and Accountability When Maintaining Automatic External Defibrillators: Infrastructuring Lifesaving Technology with an IoT Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oliver Rønn Christensen, Signe Helbo Gregers Sørensen, Anne Stouby Persson, Anne Marie Kanstrup, and Adrienne Mannov

351

373

383

Designing for the Elderly and Accessibility Improving the Language of Designing for Ageing . . . . . . . . . . . . . . . . . . . . Elena Comincioli, Alice Chirico, and Masood Masoodian Strategically Using Applied Machine Learning for Accessibility Documentation in the Built Environment . . . . . . . . . . . . . . . . . . . . . . . . . . Marvin Lange, Reuben Kirkham, and Benjamin Tannert What Happens to My Instagram Account After I Die? Re-imagining Social Media as a Commemorative Space for Remembrance and Recovery . . . . . . . Soonho Kwon, Eunsol Choi, Minseok Kim, Sunah Hwang, Dongwoo Kim, and Younah Kang

405

426

449

Education and HCI Pepper as a Storyteller: Exploring the Effect of Human vs. Robot Voice on Children’s Emotional Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Berardina De Carolis, Francesca D’Errico, and Veronica Rossano Reducing the UX Skill Gap Through Experiential Learning: Description and Initial Assessment of Collaborative Learning of Usability Experiences Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Audrey Girouard and Jin Kang What Students Do While You Are Teaching – Computer and Smartphone Use in Class and Its Implication on Learning . . . . . . . . . . . . . . . . . . . . . . . Carli Ochs and Andreas Sonderegger

471

481

501

xxxiv

Contents – Part II

Experiencing Sound and Music Technologies Experiences of Personal Sound Technologies . . . . . . . . . . . . . . . . . . . . . . . Stine S. Johansen, Peter Axel Nielsen, Kashmiri Stec, and Jesper Kjeldskov How Much is the Noise Level be Reduced? – Speech Recognition Threshold in Noise Environments Using a Parametric Speaker – . . . . . . . . . . Noko Kuratomo, Tadashi Ebihara, Naoto Wakatsuki, Koichi Mizutani, and Keiichi Zempo Stress Out: Translating Real-World Stressors into Audio-Visual Stress Cues in VR for Police Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quynh Nguyen, Emma Jaspaert, Markus Murtinger, Helmut Schrom-Feiertag, Sebastian Egger-Lampl, and Manfred Tscheligi

523

542

551

What Is Fair? Exploring the Artists’ Perspective on the Fairness of Music Streaming Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Andres Ferraro, Xavier Serra, and Christine Bauer

562

You Sound Relaxed Now – Measuring Restorative Effects from Speech Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yong Ma, Jingyi Li, Heiko Drewes, and Andreas Butz

585

Explainable AI Effects of Interactivity and Presentation on Review-Based Explanations for Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diana C. Hernandez-Bocanegra and Jürgen Ziegler

597

Human-XAI Interaction: A Review and Design Principles for Explanation User Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michael Chromik and Andreas Butz

619

Making SHAP Rap: Bridging Local and Global Insights Through Interaction and Narratives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michael Chromik

641

Quantifying the Demand for Explainability. . . . . . . . . . . . . . . . . . . . . . . . . Thomas Weber, Heinrich Hußmann, and Malin Eiband

652

Trust Indicators and Explainable AI: A Study on User Perceptions . . . . . . . . Delphine Ribes, Nicolas Henchoz, Hélène Portier, Lara Defayes, Thanh-Trung Phan, Daniel Gatica-Perez, and Andreas Sonderegger

662

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

673

COVID-19 and HCI

Addressing the Challenges of COVID-19 Social Distancing Through Passive Wi-Fi and Ubiquitous Analytics: A Real World Deployment Miguel Ribeiro1,2(B) , Nuno Nunes1,2 , Marta Ferreira1 , Jo˜ ao Nogueira1 , 2,3,4 1,2 , and Valentina Nisi Johannes Sch¨ oning 1

Instituto Superior T´ecnico, University of Lisbon, Lisbon, Portugal 2 ITI/LARSyS, Funchal, Portugal [email protected] 3 University of Bremen, Bremen, Germany 4 University of St. Gallen, Gallen, Switzerland

Abstract. During the COVID-19 pandemic, social distancing measures were employed to contain its spread. This paper describes the deployment and testing of a passive Wi-Fi scanning system to help people keep track of crowded spaces, hence comply with social distancing measures. The system is based on passive Wi-Fi sensing to detect human presence in 93 locations around a medium-sized European Touristic Island. This data is then used in website plugins and a mobile application to inform citizens and tourists about the locations’ crowdedness with real-time and historical data. To understand how people react to this type of information, we deployed online questionnaires in situ to collect user insights regarding the usefulness, safety, and privacy concerns. Results show that users considered the occupancy data reported by the system as positively related to their perception. Furthermore, the public display of this data made them feel safer while travelling and planning their commute.

Keywords: Passive Wi-Fi

1

· Survey · Mobile application · COVID-19

Introduction

Driven by the impact of the COVID-19 pandemic, governments worldwide were forced to impose lockdowns, promote work from home, and enforce other measures for citizens to keep social distancing slow-down the spread of the virus and flatten the curve to avoid overwhelming the healthcare systems. In this scenario, we saw the appearance of several technological applications emerging, ranging from symptom tracking [23,27] and pandemic planning [44] to digital contact tracing [30,31,41,43] and also reaching the field of information visualization platforms [11,46]. Especially in high-density urban areas, maintaining social distancing is intricate. After the first wave of the pandemic from February to c IFIP International Federation for Information Processing 2021  Published by Springer Nature Switzerland AG 2021 C. Ardito et al. (Eds.): INTERACT 2021, LNCS 12933, pp. 3–24, 2021. https://doi.org/10.1007/978-3-030-85616-8_1

4

M. Ribeiro et al.

May 2020, many countries in the northern hemisphere saw a decrease in infection rates leading to a return of many economic activities. Driven by the need to foster economic growth, many countries incentivized unconfinement measures, from mandatory use of masks and sanitation to enforcing capacity limitations in public spaces. This paper describes a technology-based measure implemented in a touristic European Island after the first wave of the pandemic. The system evolved from an existing community based passive Wi-Fi infrastructure deployed as part of a tourism flow sensing research project [29]. The infrastructure uses a network of Wi-Fi routers to passively detect the presence of wireless-enabled devices in their range [36]. Through machine learning, the infrastructure estimates the presence and flow of people both historically and in real-time. Unlike existing systems (e.g., Google popular times), the system is open to the local community enabling the deployment of additional services, including ubiquitous analytics [39] adapted to the local context. Here, we describe how we adapted the original infrastructure to help citizens and visitors, including local businesses and health and tourism authorities, leveraging open information about people’s presence and flow. The infrastructure was deployed on 93 specific points of interest (POIs), emphasizing touristic places and popular public locations like shopping centers. Sensing human presence has been approached in several manners, from manual people counting, surveys to wearable trackers [7]. In the context of the pandemic, monitoring mobility at scale and in real-time was an important requirement. Several methods were attempted, from video processing [15,38], crowdsourcing smartphone applications [33], cell tower and call record, and non-intrusive wireless monitoring (NIWM) [5,21,37,42]. While most techniques require user knowledge or intervention in the data acquisition process, NIWM does not. Either via cell signals, Bluetooth, or Wi-Fi devices, the number of people in specific locations can be sensed automatically and anonymously using NIWM. However, this data is hard to validate, as users often do not know they are being monitored or are not actively involved in the data collection process. To overcome this challenge, the system described here collects human presence data anonymously and presents the results back to citizens transparent and accessible. In this paper, we present our design approach and discuss how users understand and react to this real-time information. We contribute to a deeper understanding of such a system’s impact in helping people maintain social distancing and the broader implications in terms of accuracy of the methods employed and the safety and privacy concerns that such information could entail. 1.1

Contributions and Research Questions

The work described in this paper was the response to the regional call to action to help the community return to a “new normal”. The passive Wi-Fi infrastructure already in place in the island [29] was capable of estimating the presence and flow of people in public spaces. The existing system provided an opportunity to adapt the information for citizens and tourists, to help them make informed decisions when planning and moving through public locations. The system was

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

5

then adapted from a generic passerby counter to a system to reinforce the social distancing measures of COVID-19 situation in the following ways: i) increase the existing network of passive Wi-Fi sensors to 93 POIs covering not only touristic POIs but also essential public places such as shopping centers and squares; ii) developing a mobile application that could show this information to citizens and locals, together with COVID-19 safety measures and tourist information about POIs and attractions; and finally, iii) deploying an in situ survey to present the occupancy data and probe passerby about the accuracy of the displayed data, as well as the sense of safety, privacy, and satisfaction of the location. Through this research, the authors intend to answer the following research questions: RQ1: Can we validate passive Wi-Fi data as a representation of occupancy of a POI for COVID-19 occupancy monitoring? RQ2: Can occupancy data automatically collected via passive Wi-Fi increase the sense of safety and crowd awareness in public locations? RQ3: Can we use this information to help citizens make informed decisions about when to visit public locations based on their occupation?

2

Related Work

This research is primarily motivated by two areas of prior work, combining non intrusive people sensing and data presentation to citizens to collect feedback. The area of people sensing has had in the past the attention of researchers mainly due to the unnecessary intervention of the people being sensed. The technologies and techniques vary and aim to target the masses with automated mechanisms that require little to no human intervention. Examples of this are found in video surveillance analysis, where the video feed of existing surveillance cameras can be processed to enable head counts by using convolutional neural networks to overcome the challenges of low resolution images and poor lighting [15]. Recently a study [38] was done using thermal imagery to tag people for COVID-19 with signs like high body temperature by automatically detecting the faces and the temperature of the people passing in front of a thermal camera. Other efforts were done using big data of cell phone records [4,40] to detect mobility traces, interaction patterns and taking advantage of functionalities such as handover and cell (re)selection, which are used to maintain seamless coverage for mobile end-user equipment between cells. The study by Alsaeedy and Chong [2] uses those parameters to identify the regions at risk of infections with high human mobility by using the existing cellular network. Also, Ahmed in [1] proposed a similar framework to control the spread of COVID-19 using the cellular system by using the mobile identifiers of infected people and calculating the distances of other phones nearby in those cell towers to deliver that information to the clinical centers for close contacts of suspects. Wi-Fi has been used in the pre-pandemic to detect gatherings and passerby, with studies being done in football games [5], universities campuses [3] and hospitals [37]. The motivations of many of these studies are diverse, some look at energy waste on scanning methods [3], realistic facility management and planning [37], while others looked at crowding factors, flock detection and waiting

6

M. Ribeiro et al.

times, speed and frequent paths [21] and even social information like popularity of events (in the case of [4] singers in concerts). In the pandemic context, Wi-Fi has also been used to detect the contagion risk in public transport [10] by using multiple sensors that monitor the air quality, temperature humidity and use WiFi probes to detect the people present thus creating an index for contagion risk. Another solution that employs Wi-Fi probe requests is presented in [42] with the goal to estimate indoor crowd densities based on dynamic fingerprints based on signal strength and proposing a crowd density estimation solution based on the positioning algorithm. The authors made experiments in a laboratory class and in three public social activities in a footbridge, railway and subway demonstrating how this can be used to estimate crowd density. Wi-fi was also proposed in a framework to monitor vital respiratory signs of patients by measuring the Wi-Fi frames reflected by the human body standing between a dedicated emitter and receiver and applying bandpass filtering to extract respiratory components of the patients [25]. 2.1

Feedback via Voluntary Walk-Up Systems

While user surveys and online forms are viable ways for citizens to contribute with their opinions, the idea of gathering this type of information in situ without the need for a human interviewer opens opportunities for new studies. HCI brings new possibilities of presenting and gathering information from the users by providing interactive and attractive interfaces to grasp the passerby’s attention often done via public displays. A recurrent problem with this method is that the users are often presented with automatically sensed information, that may not always represent the reality, and the user may feel the need to give an opinion or to even challenge the presented data. A classic example of this is the presentation of real-feel weather conditions or environmental parameters. The area of surveying is essential to gather the citizens perspectives about different subjects of our society and products. Several approaches have been done to assess user satisfaction or inquiry through interactive technologies [6,19,20, 28]. One of the studies done in this area presents Roam.IO [18], which proposes a hybrid data approach combining data from IoT sensing with qualitative data contributed by citizens. The authors developed a public installation consisting of a tangible robot-like interface placed in passageways, to entice and encourage the public to walk-up and answer questions. The users are questioned to suggest what the data might represent and enrich it with subjective observations with success in user engagement with this apparatus. 2.2

Citizen Science

Several attempts have tried to tackle the problem of gathering data through citizens in ubiquitous ways [17,45]. This led to a range of interactive technologies, focused on human-centred data collection. An example is VoiceYourView [45] which uses speech recognition and natural language processing to provide feedback of how citizens perceive and feel about the public spaces around them.

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

7

Physical tangible mechanisms have also been used for public opinion gathering such as polling systems to urban voting and visualizations [24]. Another tangible approach was EmoBall [12] and Mood Squeezer [14] which allows for playfulness in office buildings where a lightweight technology was designed to ask people to reflect on their mood by squeezing a colored ball from a box set. VoxBox [16] presented a tangible system for gathering opinions in situ at an event through playful and engaging interaction via a large panel of interaction turning the questionnaire itself into a toy that made people want to fidget with it. CommunitySourcing [17] was also used to investigate the potential of community sourcing by designing, implementing and evaluating a vending machine that allowed users to earn credits by performing tasks using a touchscreen attached to the machine and physical rewards were dispensed through traditional vending mechanics. Similarly, SmallTalk [13] explored how tangible devices used as a toy with spinners, dials and buttons can be used to gather feedback from children in the context of a theater performance via a tangible survey system to capture what children thought of the performance they had seen. Recently Diethei [8] explored socio-psychological processes and motivations to share personal data during a pandemic.

3

Data and Deployment

This system’s main components rely on the presence and flow sensing infrastructure, providing user information via dashboards, web plugins, a mobile app, and finally collecting in situ feedback. The sensing infrastructure is based on a network of passive Wi-Fi routers described below. This infrastructure then powers the two other components by enabling data to be processed and presented to the end-users. The presence and flow data is presented to different stakeholders in different formats, from a global dashboard mostly targeting authorities to web plugins (that can be included in websites) targeting local businesses, thus informing them of people’s presence at different times. Then, the data is consumed by an official Tourism Board app that presents the list of safe locations to visit and at what times. Finally, the surveys are generated as interactive webpages that can be deployed on tablets in situ or accessed by mobile phones via a QR code. The survey is used to collect feedback from the users of those locations about the accuracy of the data, safety of the location and overall satisfaction. 3.1

Passive Wi-Fi Data Gathering

The presence data presented in the visualizations is gathered from a passive Wi-Fi platform [29,36]. This infrastructure was originally designed to support a community-based Wi-Fi tracking system to understand mobility at scale. The original system provided several interactive dashboards [32,34,35] to help communities easily run systematic analysis of tourists’ mobility patterns in the destinations, contributing in new ways in visualizing spatio-temporal mobility data. This system is composed of Wi-Fi routers spread across all parishes of the region, as shown in Fig. 1. The data is collected in real-time by Wi-Fi routers

8

M. Ribeiro et al.

Fig. 1. Map of Madeira Islands (Portugal) showing the locations of the passive Wi-Fi routers deployed.

spread across multiple locations covering all regions’ parishes. The data collected and sent to a central server includes a location identifier, a timestamp and an anonymous identifier (based on the MAC address) of each Wi-Fi enabled device present in a radius of approximately 80 m from the router (depending on the location and obstacles). The processed results are made available via an API that allows developing standalone implementations, thus providing real-time information about the number of devices detected in the last hours, location meta-data, and historical hourly data that serves as a comparison to the current counts. The locations were chosen based upon common points of interest selected in coordination with the local tourism bureau and include governmental buildings close to the street, tourism offices, plazas, restaurants, cafes, and viewpoints where the routers were installed in the nearest facility. The routers placed in each location needed an electrical connection and internet access, thus posing challenges in terms of coverage and location. They had to be placed indoors close to outlets and on a suitable site provided by the existing building. A challenge in adapting the infrastructure for COVID-19 security capacity was asking each site to provide the maximum safety number for each space. To overcome both challenges, the research team worked closely with local authorities to find and test a reasonable match between the router’s range and the spaces’ capacity. This process was tested initially in a private beach club, which provided very good conditions to evaluate the sensing infrastructure’s accuracy. The club was only accessible via a controlled entrance and provided daily counts of admissions per hour and the maximum number of people inside the facilities on any given hourly period. This data was used as ground truth for several weeks to evaluate the accuracy of the machine learning methods used to estimate both the number of entrances and the total number of people inside the facility from the number of devices detected by the router. The ground truth facility had an overall capacity of 500 people and it was monitored with a single

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

9

router located in the main entrance building. While this shows the simplicity and ease of deployment of the sensing infrastructure, not all the POIs could rely on the same level of accuracy since, in many places, there was no access control or no structured capacity defined. In these cases, the research team judged from inspection of maps and local screening based on signal strength measures. Finally, in some POIs, it was impossible to make the assessment, and hence the research team resorted to 60% of the maximum pre-pandemic capacity. 3.2

Dashboard and Web Plugins

Motivated by the rapid spread of COVID-19, and inspired by the social sensing paradigm, the contributions of this section lie in the implementation of an interactive platform based on the Wi-Fi tracking infrastructure, to visualize information related to this enriched dataset of spatio-temporal data. In particular, we here present the visualizations related to the COVID-19 disease diffusion and its effect on human mobility in the islands and describe the interactive data visualization application. We designed and implemented a web-based interactive data visualization built using Web technologies, focusing on the visualizations related to the COVID-19 period, showing the data in percentages of the total for each location, in a way to preserve the counts of some routers that are located in private infrastructures. These data representations have been designed to present the human presence considering the period when the pandemic of COVID-19 was experienced in Portugal. In particular, as it is possible to see in Fig. 2, the presented data show the percentage of users in a specific area, in comparison with the last week before the day the state of emergency was declared in Portugal and the first case in Madeira Islands. The dashboard presents an interactive visualization of human presence data during the COVID-19 outbreak, and to compare this data with the patterns of previous weeks. The visualized data is collected exploiting social sensing, and in particular, a community-based passive Wi-Fi tracking infrastructure, deployed in the Madeira Archipelago.

Fig. 2. (Left): Presents the information grouped by municipality; (Right) Zoomed view presents the ungrouped information focused on a specific area.

10

M. Ribeiro et al.

In addition to this, a web widget was also created being able to be embedded in websites, (which was also used for the in situ survey shown in Fig. 3) so that each institution that has a router in its facilities can also inform its visitors about the current status and historical trend for each hour. This widget is currently used in the Private Beach Club website informing their website visitors of their affluence in real-time. 3.3

Mobile Application

The final touchpoint consisted of a mobile application designed for the Health and Tourism authorities responsible for implementing the Islands’ safe tourism system. This system relied on the airport (and port) PCR COVID-19 testing. The visitors could either upload the results of a COVID-19 PCR test to the app and then pass the health check at the airports’ arrivals. Alternatively, they could perform a free PCR test on arrival and receive the test results via email. During the 14-day monitoring period, the visitors were also asked to fill out a voluntary epidemiological survey.

Fig. 3. Mobile application screens: a) Experiences - This view gives the user an overview of the current points collected and experiences exchange for points, as well as the history in a vertical chronological view; b) How I Feel - Shows the common user recommendations from the local health authorities and can call emergency numbers; c) Plan - This is where the user can explore the region in a map view and visualize the real-time occupancy of the locations marked by a shield that varies in color. Upon touching a site, the user is presented with historical occupancy for the different weekdays’ data and shared information about the place, such as website, address, contacts, and schedule; d) Privacy - In this tab allows the user to view in a transparent manner how the information that is being presented is collected, and the ability to set notification permissions and geolocation (used to locate the user in the map).

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

11

Thus, the app was part of a more extensive safety system, which included testing all visitors, asking them to fill out an epidemiological survey and a daily follow up of COVID-19 related symptoms. To nudge visitors to perform the PCR test before arrival and as a way to incentivize daily epidemiological monitoring, the app also implemented a gamification mechanism that awarded users points each time they performed a COVID-19 safe activity. Although the app’s gamification component is beyond this paper’s scope, it enabled a user to accumulate points by completing tasks and challenges of visiting safe locations. The points could then be exchanged for free experiences proposed by the local tourism office. The application interface starts by presenting the user with opening instructions for the first use, which helps the user navigate through the four main screens of the application and understand the points system, the challenges, and mechanics. Then the information is depicted in Fig. 3 and categorized into four tabs. Both the real-time map data and the historical occupation charts are powered by the passive Wi-Fi infrastructure that collects data anonymously in these locations. 3.4

In Situ Survey

The in situ surveys were deployed in a total of four locations (described in more detail in Table 2). These surveys were used to present and collect information in selected POIs and survey the passerby’s about the sense of safety and satisfaction of the location, and inquiry about privacy and safety concerns. This method also provided a way to crowdsource the accuracy of the information depicted, asking passersby to confirm the estimate and fine-tune the machine learning algorithms and the overall range and occupancy estimates. The user interface was developed using web technologies suitable for responsive screen sizes and were deployed in a tablet to be answered via smartphone. The interface was designed as follows: – To show the user the current location and current count (the system counts the number of devices in the Wi-Fi range and estimates the number of people based on a custom made algorithm) via a widget with an animated waveform gauge. The animation is supposed to draw attention to the screen (see Fig. 4, top left); – Based on the real-time Wi-Fi counts, the interface shows an average hourly count history with the current hour highlighted (Fig. 4 Top); – The interface shows the interactive survey. The survey gathers information about the users feeling about the occupancy levels of the location (as reported by the system interface), their safety perception, and the relevance of the realtime widget regarding their perceptions; – The interface cycles through question panels when no interaction is recorded for 15 s. The cycling of the panels is designed to draw attention to the interactive screen; – The system sends data to the database, when the user: answers all questions or does not interact for 15 s after having answered a question.

12

M. Ribeiro et al.

Fig. 4. In situ interactive survey Interface components diagram.

The survey had the option of two languages (English and Portuguese), has a QR-Code for the user to scan and answer the survey on their mobile phone thus avoiding physical touch with the tablet and has an “Info” button to show the project details and information about how the data is gathered and related privacy issues. The survey questions are organised according to three kinds of answers, each one depicted by a specific icon: a star rating to rate the place interest, a shield to rate safety and a smiley for accuracy and trustworthiness score on a 5 point Likert scale We kept the number of questions low and succinct to maximize user engagement and a swift interaction with minimum effort. The questions are presented in Table 1, grouped into two tabs of questions that cycled between them. Table 1. In situ survey questions #

Question

Rating scale Pannel Connotation

Q1 Do you find this place interesting?

Star

1

Positive

Q2 Do you find this place COVID-19 safe?

Shield

1

Positive

Q3 Do you find this information accurate?

Smiley

1

Positive

Q4 Do you find this information trustworthy?

Smiley

1

Positive

Q5 Do you find this information useful?

Smiley

2

Positive

Q6 Do you find this information intrusive?

Smiley

2

Negative

Q7 Do you find this information improves safety? Smiley

2

Positive

Q8 Do you find this information violates privacy? Smiley

2

Negative

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

3.5

13

Deployment

The in situ survey was deployed in four POIs and took place during the months of September to December of 2020. The locations were the research institute cafe (where it was piloted for a week), a private beach club, the gift shop of one of the most popular tourist attractions (viewpoint) and a Hotel lobby. In all of these places a tablet was used on a stand to show the relevant information and prompt the users with the questions. The locations, described in Table 2 were chosen for their characteristics of being locations that drew the attention of locals and tourists, and for their facilities that allowed an appreciate placement of the tablets indoors and in secure yet accessible spots in the premises. Table 2. Location details Type

Location

Research Institute

Restaurant/Bar n/a

Limit Nr Answers Nr days 77

6

Touristic attraction (Viewpoint) Gift Shop

n/a

28

128

Beach club

Entrance

500

90

131

Hotel

Reception

n/a

35

105

230

–

Total

4

Results

In this section, we will present the results from the in situ survey, relative to how the users classified the information provided and perceived their safety at the locations. Besides, we present the results of modeling the Wi-Fi counts with ground truth data in one site to estimate the real number of people present. 4.1

Passive Wi-Fi Ground Truth Validation

One of the four sites where the in situ survey was deployed, the private beach club, was used to provide ground truth to validate the data collected from the rest of the nodes of the passive Wi-Fi system and remodel the data to estimate the real number of people in the premises. To support the unconfinement of people after the first wave of COVID-19, the regional authorities asked all public sites to limit occupancy. For this specific site, the number of people inside the premises was limited to 500. Since the access was controlled electronically, this site provided unique conditions to evaluate the accuracy of the Wi-Fi sensor. The ground truth data corresponds to 356 data points covering 44 days between June and August. The whole premise was covered by a single Wi-Fi sensor placed in the main building after the entrance. While the Wi-Fi counts help to show the peak times of occupation and the overall curve along the days of the week, those numbers are not exact counts.

14

M. Ribeiro et al.

Fig. 5. Weekday and hourly correlations between the Wi-Fi counts and ground truth.

This is mainly due to people out of the Wi-Fi range having their wireless turned off, among other factors. To estimate the real number of people in the premises from the real-time Wi-Fi counts, we compared and created data models with the official hourly number of entrances and exits controlled via the electronic access control system. The time slot used for the comparisons was 60 min. For each hour and weekday, we computed the correlation between these variables. We tried to estimate the real number of people through data modeling. The data gathered in the private beach club shows strong Persons’ correlations with the ground truth data. The correlation for the dataset is 0.64, and Fig. 5 shows the breakdown in weekdays and hours. In this analysis, each correlation is done using 6 data points. The low values in some isolated hours correspond to the beginning and end of the day, where abnormal movements may have occurred during the procedures of opening and closing of the facilities that did not provide correlations as good as the remaining hours. We also attempted to model the number of actual people entering and exiting the premises against the router counts with several regression methods. The features were extracted by grouping the data hourly for each weekday. Each row represented one hour of data on a specific weekday from the location. The number of hours of ground truth varied, as also did the closing times. The total

Fig. 6. Evolution of R2 (left) and explained variance (right) with incremental steps of train/test percentages for seven methods.

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

15

Fig. 7. Beach club ground truth counts vs router counts.

Fig. 8. Modeled router counts of the beach club to fit the ground truth data.

number of days was 44, ranging from 2020-06-15 to 2020-07-29. This resulted in 590 data points with the following features: hour, weekday, count. The Scikitlearn python library was used to test the following common methods used in regression: – – – – –

Decision Tree Regressor (DTR) K-Neighbors Regressor (KNR) − k = 100 Random Forest Regressor (RFR) − Estimators = 100 Extra Trees Regressor (ETR) − Estimators = 100 Gradient Boosting Regressor (BGR) − Learning Rate = 0.1, Estimators = 100 – Linear Support Vector Regressor (LSVR) = Tolerance = 1e−4 – Epsilon Support Vector Regressor (SVR) − Degree = 3, Epsilon = 0.1

The parameter sweep was done for the train/test ratio, ranging from 10% to 90%, with step increments of 10%. These classifiers were run 1000 times for each of the steps and the average accuracy for each classifier was taken for the results shown in Fig. 6. The method that shows higher results when using the same dataset for trains and test (train size = 1) is the Gradient Boost Regression. Using this method, we plotted in Fig. 7 the original data points of router counts and ground truth, and in Fig. 8 the regression made using the Random Forest Regressor, with values of R2 and explained variance of 0.92. These results show that the routers data can be reliably used to predict the real number of passersby in certain locations by modeling for each location, or location typology. The accuracy depends on how the router is installed, its range,

16

M. Ribeiro et al.

Table 3. Statistical analysis of replies (Q1-Interesting; Q2-Safe; Q3-Accurate; Q4Trustworthy; Q5-Useful; Q6-Intrusive; Q7-Improves safety; Q8-Violates Privacy) Parameter Q1

Q2

Q3

Q4

Q5

Q6

Q7

Q8

N

185 182 174 163 172 155 155 149

Average

4.58 4.41 4.17 4.26 4.40 3.46 4.43 3.29

Std. Dev

0.95 1.09 1.19 1.13 1.08 1.60 1.01 1.69

Median

5

5

5

5

5

4

5

4

how many people leave Wi-Fi on, as well as being influenced by the random MAC addresses, however, as we can see, it is still possible to create accurate models that predict well the occupancy of a location. 4.2

In Situ Survey

During the survey deployment, we received a total of 230 responses. All of the questions were formulated affirmatively except two (Q6 and Q8), concerning privacy and intrusiveness. The user’s answers can be interpreted as the higher the scales’ rating, the better. The statistics for the 230 replies to the eight questions are shown in Table 3. It shows that Q6(privacy) and Q8(intrusiveness) have more spread out standard deviations, while the remaining questions scored high averages, showing that the public information derived from the passive Wi-Fi data was correlated with the location’s perceived occupancy. In general the users reported interest in the site, and felt COVID-19 safe. Since the survey could be abandoned without answering all of the questions, the total number of users that replied to all questions was 130. The answers regarding interest in the place (Q1), perceived safety in the place, accuracy of info, trustworthiness, usefulness, and improved safety of the place through the info provided (Q2, Q3, Q4, Q5, Q7), are all

Fig. 9. Average question results across the different locations.

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

17

Fig. 10. Number of questions answered by Wi-Fi count(left) and percentage shown in the widget(right).

quite positive (between 4 and 5 Likert scale rating). The questions regarding intrusiveness of the information (Q6) and violation of privacy (Q8) recorded lower scores, compared to the previous answers, but still quite high, (between 3 and 4), according to the Likert scale. Since the questions were formulated in a negative form, high scores in these two questions indicate users perceptions of the information being intrusive and privacy violation. Figure 9 shows the score distribution of the questions across the four locations is consistent across all sites, except for questions Q6 and Q8 regarding privacy and intrusiveness. For these questions, the Hotel and Bar show users being more concerned about privacy (Q8) and intrusiveness (Q6). To note that in all questions, the maximum standard deviation (SD) was 1.69 and only 2 questions had SDs above 1.19 in a 5 point Likert scale, which implies that most answers in these 2 questions ranged between ∼2.4 points of the scale, thus not being able to extract conclusions as strong and consistent as the remaining questions. In Fig. 10 we present the number of questions answered compared to the number of Wi-Fi devices detected in the past 60 min to the reply and the occupation percentage shown in the widget. The results reveal no direct increase or decrease resulting from the percentage shown and the number of questions answered. Six, was the most prevalent number of questions answered when the location showed more occupancy.

Fig. 11. Distribution of number of questions answered by each user.

18

M. Ribeiro et al.

Fig. 12. Number of daily responses throughout the deployment (Broken Y axis).

Figure 11 shows the distribution of the number of questions answered by each user, revealing that the vast majority of users replied to all eight questions or just one single question before abandoning the survey. The deployment spanned from September to December of 2020. The number of daily responses varied considerably, as shown in Fig. 12, with a higher frequency happening in the initial days and then rapidly fading out through the deployment. To note that across the four locations, at least one response was recorded every day. We also plotted the replies for questions 1–5 and 7 against the percentage shown in the widget in Fig. 13. These were the questions with positive connotation aimed to understand the user’s satisfaction with the location and the perception of accuracy of the data being presented. Lastly, the language in which the questionnaires were answered may differentiate the users that answered from local and foreigners. The results presented in Fig. 14 (left) shows that the majority of survey entries were done in Portuguese, and that division is maintained across the users of the other locations, with the exception of the hotel reception, in which the majority of entries were done in English as seen in Fig. 14 (right).

Fig. 13. Average reply for questions 1–5 and 7 (positive connotation) against the percentage shown in the widget.

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

19

Fig. 14. Number of survey entries per language (left) and number of questions answered per survey language (right).

5

Discussion

As a low-cost technological solution to help unconfinement measures during the pandemic, the infrastructure described here provided some useful results and discussion points. Monitoring and modeling people’s behavior is always a sensitive issue. The passive Wi-Fi approach is both anonymous and practical for providing real-time information about the people’s presence, occupancy and flow. Previous research showed this method was effective in detecting people’s presence and flow while preserving privacy [29,36]. Although Wi-Fi devices expose IDs that could potentially be used to track the owners, our dataset clearly shows that this is impractical with the high number (96%) of randomized IDs exposed in recent iOS and Android implementations. Although, to increase privacy, the system does not store device IDs, our past data shows the randomization percentage has risen significantly in the last years (56% since 2016). This is an important privacy by design feature, making passive Wi-Fi more trustworthy than other alternatives, e.g., using cameras [47] or access control systems or even active Wi-Fi logging [26]. Although systems like Google popular times provide similar information, these datasets are not publicly available and therefore, are not a practical solution for communities such as those described here. As drawbacks, the passive Wi-Fi system requires an infrastructure to operate (power, internet connection, backend for ML algorithms and provide real-time and historical data). While our results show a strong correlation between the passive Wi-Fi data and the ground truth, the system is naturally prone to inaccuracies. The location (including obstacles), the limited range, and dependency on having Wi-Fi enabled can naturally influence the system’s accuracy. One could argue that such potential inaccuracies are a built-in feature as they end up increasing privacy protection (e.g., acting as a non-intentional mechanism to implement differential privacy [9]). Despite the limitations, our results clearly show that passive Wi-Fi is a viable alternative for capacity monitoring in the wild, answering RQ1. Besides, the in situ surveys were a mechanism to understand people’s reaction of this information. First, the survey results confirmed that users perceived the occupancy data displayed to be quite accurate, confirming from a user perspective our RQ1 (Q3–Q5). The results show that users were overall pleased with the information provided. From the answers regarding Q7 (safety improval) the average result of 4.43 shows that the users perceived that the system presented

20

M. Ribeiro et al.

improved the citizens feeling of safety in that location while also finding the place COVID-safe (Q2), which also comes from a combination of the measures taken by the local authorities. Secondly, our deployment confirms that in situ surveying is prone to the novelty effect, exciting users in the beginning, while not supporting long term engagement of the same number of users [22]. However, it did continue to engage some users over the whole deployment (4 months) since we recorded at least one answer per location over the period of 127 days. Finally, concerning privacy and intrusiveness (Q6 and Q8), the users express some negative concerns about the system as the Likert’s scores are quite high (3 to 4) and the question phrased negatively. Further probing of the users would be necessary to understand their concerns in more details. Future work should address differences between in situ and mobile app use of this same information. The mobile application has been released, being promoted mainly via touristic channels and was installed by over 9000 users in Google and Apple stores (Jan 2021 to Apr 2021). From the usage data, and the time spend in each screen, the users explored on average 13 locations of the interactive map and opened the hourly occupation charts, thus showing that the application was used to inform users of the real-time occupation in several locations. In addition, given the positive feedback of the data accuracy (Q3) and trustworthiness (Q4) from the in situ survey (the same rates shown in the app), the data shown in the mobile application enabled the users to make informed decisions about their commute plans around the region as per RQ3.

6

Conclusion

With this work, describe and evaluate the deployment of three components that aim to inform citizens about the real-time occupation of public spaces by using a passive Wi-Fi system. The challenge of estimating the real number of people from the Wi-Fi counts was modeled for a use case with accurate ground truth, that provides a solid base for modeling data accurately. The data modeling from the use case used for modeling here set the ground to update the rest of the WiFi locations to be used in that way, opening the possibility of creating models for locations with similar characteristics once the basic info is provided, such as the capacity of the places, and the accuracy of how much of the passerby does the system capture, while adjusting the routers so to cover all ground necessary to estimate the number of people in that location. The positive feedback of the survey system encourages future work to turn the web dashboard presented into a larger public display to be placed in the cities and inform citizens in real time about the presence of people in locations nearby with the updated models for each location. While the mobile application will also be updated to include quick user surveys that prompt users with questions about the experience of using the application and the general feel of the visited locations as well as feedback about the presented data.

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

21

References 1. Ahmed, A.M.: Designing a framework to control the spread of COVID-19 by utilizing cellular system. Kurdistan J. Appl. Res. 5, 146–153 (2020). https://doi.org/ 10.24017/covid.16 2. Alsaeedy, A.A.R., Chong, E.K.P.: Detecting regions at risk for spreading COVID19 using existing cellular wireless network functionalities. IEEE Open J. Eng. Med. Biol. 1, 187–189 (2020). https://doi.org/10.1109/OJEMB.2020.3002447. Conference Name: IEEE Open Journal of Engineering in Medicine and Biology 3. Baniukevic, A., Jensen, C., Lu, H.: Hybrid indoor positioning with Wi-Fi and bluetooth: architecture and performance. In: 2013 IEEE 14th International Conference on Mobile Data Management (MDM), vol. 1, pp. 207–216 (2013). https://doi.org/ 10.1109/MDM.2013.30 4. B¨ ohmer, M., Hecht, B., Sch¨ oning, J., Kr¨ uger, A., Bauer, G.: Falling asleep with angry birds, Facebook and kindle: a large scale study on mobile application usage. Presented at the (2011) 5. Bonn´e, B., Barzan, A., Quax, P., Lamotte, W.: WiFiPi: involuntary tracking of visitors at mass events. Presented at the (2013). https://doi.org/10.1109/WoWMoM. 2013.6583443 6. Brignull, H., Rogers, Y.: Enticing people to interact with large public displays in public spaces. Interact 3, 17–24 (2003) 7. Buettner, M., Prasad, R., Philipose, M., Wetherall, D.: Recognizing daily activities with RFID-based sensors. Presented at the (2009). https://doi.org/10.1145/ 1620545.1620553 8. Diethei, D., Niess, J., Stellmacher, C., Stefanidi, E., Sch¨ oning, J.: Sharing heartbeats: motivations of citizen scientists in times of crises. Association for Computing Machinery, New York (2021). https://doi.org/10.1145/3411764.3445665 9. Dwork, C.: Differential privacy: a survey of results. In: Agrawal, M., Du, D., Duan, Z., Li, A. (eds.) TAMC 2008. LNCS, vol. 4978, pp. 1–19. Springer, Heidelberg (2008). https://doi.org/10.1007/978-3-540-79228-4 1 ˙ G¨ ˇ Ya˘ 10. Fatih, S ¸ I, okhan, A.N., Pani´c, S., Stefanovi´c, C, gano˘ glu, M., Prilinˇcevi´c, B.: Covid-19 risk assessment in public transport using ambient sensor data and wireless communications. Bull. Nat. Sci. Res. 10(2), 43–50 (2020). https://doi.org/10.5937/ bnsr10-29239 11. Florez, H., Singh, S.: Online dashboard and data analysis approach for assessing COVID-19 case and death data. F1000Research, vol. 9, no. 570, p. 10. 2020.12688/f1000research.24164.1 12. Fuentes, C., Rodr´ıguez, I., Herskovic, V.: EmoBall: a study on a tangible interface to self-report emotional information considering digital competences. In: Bravo, J., Herv´ as, R., Villarreal, V. (eds.) AmIHEALTH 2015. LNCS, vol. 9456, pp. 189–200. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-26508-7 19 13. Gallacher, S., Golsteijn, C., Rogers, Y., Capra, L., Eustace, S.: SmallTalk: using tangible interactions to gather feedback from children. Presented at the (2016). https://doi.org/10.1145/2839462.2839481 14. Gallacher, S., et al.: Mood squeezer: lightening up the workplace through playful and lightweight interactions. Presented at the (2015). https://doi.org/10.1145/ 2675133.2675170

22

M. Ribeiro et al.

15. Gao, C., Li, P., Zhang, Y., Liu, J., Wang, L.: People counting based on head detection combining Adaboost and CNN in crowded surveillance environment. Neurocomputing 208, 108–116 (2016). https://doi.org/10.1016/j.neucom.2016.01.097 16. Golsteijn, C., et al.: VoxBox: a tangible machine that gathers opinions from the public at events. Presented at the (2015). https://doi.org/10.1145/2677199. 2680588 17. Heimerl, K., Gawalt, B., Chen, K., Parikh, T., Hartmann, B.: CommunitySourcing: engaging local crowds to perform expert work via physical kiosks. Presented at the (2012). https://doi.org/10.1145/2207676.2208619 18. Houben, S., et al.: Roam-IO: engaging with people tracking data through an interactive physical data installation. In: Proceedings of the 2019 on Designing Interactive Systems Conference (DIS 2019), pp. 1157–1169. Association for Computing Machinery (2019). https://doi.org/10.1145/3322276.3322303 19. Houben, S., Weichel, C.: Overcoming interaction blindness through curiosity objects. In: CHI ’13 Extended Abstracts on Human Factors in Computing Systems (CHI EA 2013), pp. 1539–1544. Association for Computing Machinery (2013). https://doi.org/10.1145/2468356.2468631 20. Ju, W., Sirkin, D.: Animate objects: how physical motion encourages public interaction. In: Ploug, T., Hasle, P., Oinas-Kukkonen, H. (eds.) PERSUASIVE 2010. LNCS, vol. 6137, pp. 40–51. Springer, Heidelberg (2010). https://doi.org/10.1007/ 978-3-642-13226-1 6 21. Kjærgaard, M.B., Wirz, M., Roggen, D., Tr¨ oster, G.: Mobile sensing of pedestrian flocks in indoor environments using WiFi signals. Presented at the (2012). https:// doi.org/10.1109/PerCom.2012.6199854 22. Koch, M., von Luck, K., Schwarzer, J., Draheim, S.: The novelty effect in large display deployments-experiences and lessons-learned for evaluating prototypes. In: Proceedings of 16th European Conference on Computer-Supported Cooperative Work-Exploratory Papers. European Society for Socially Embedded Technologies (EUSSET) (2018). https://doi.org/10.18420/ecscw2018 3 23. Koehlmoos, T.P., Janvrin, M.L., Korona-Bailey, J., Madsen, C., Sturdivant, R.: COVID-19 self-reported symptom tracking programs in the united states: framework synthesis. J. Med. Internet Res. 22(10), e23297 (2020). https://doi.org/10. 2196/23297 24. Koeman, L., Kalnikait´e, V., Rogers, Y.: “Everyone is talking about it!”: a distributed approach to urban voting technology and visualisations. Presented at the (2015). https://doi.org/10.1145/2702123.2702263 25. Li, F., Valero, M., Shahriar, H., Khan, R.A., Ahamed, S.I.: Wi-COVID: a COVID19 symptom detection and patient monitoring framework using WiFi. Smart Health 19, 100147 (2021). https://doi.org/10.1016/j.smhl.2020.100147 26. Meneses, F., Moreira, A.: Large scale movement analysis from WiFi based location data. Presented at the (2012). https://doi.org/10.1109/IPIN.2012.6418885 27. Menni, C., et al.: Real-time tracking of self-reported symptoms to predict potential COVID-19. Nat. Med. 26(7), 1037–1040 (2020). https://doi.org/10.1038/s41591020-0916-2 28. M¨ uller, J., Alt, F., Michelis, D., Schmidt, A.: Requirements and design space for interactive public displays. Presented at the (2010). https://doi.org/10.1145/ 1873951.1874203 29. Nunes, N., Ribeiro, M., Prandi, C., Nisi, V.: Beanstalk: a community based passive Wi-Fi tracking system for analysing tourism dynamics. Presented at the (2017). https://doi.org/10.1145/3102113.3102142

Addressing Challenges of COVID-19 Social Distancing with Passive Wi-Fi

23

30. Oswald, M., Grace, J.: The COVID-19 contact tracing app in England and ‘experimental proportionality’. SSRN Electron. J. (2020). https://doi.org/10.2139/ssrn. 3632870 31. Park, Y.J., et al.: COVID-19 national emergency response center, epidemiology and case management team: contact tracing during coronavirus disease outbreak, South Korea, 2020. Emerg. Infect. Dis. 26(10), 2465–2468 (2020). https://doi.org/ 10.3201/eid2610.201315 32. Prandi, C., Nisi, V., Ribeiro, M., Nunes, N.: Sensing and making sense of tourism flows and urban data to foster sustainability awareness: a real-world experience. J. Big Data 8(1), 1–25 (2021). https://doi.org/10.1186/s40537-021-00442-w 33. Ram, N., Gray, D.: Mass surveillance in the age of COVID-19. J. Law Biosci. 7, lsaa023 (2020). https://doi.org/10.1093/jlb/lsaa023 34. Redin, D., Vilela, D., Nunes, N., Ribeiro, M., Prandi, C.: ViTFlow: a platform to visualize tourists flows in a rich interactive map-based interface, pp. 1–2. IEEE (2017). https://doi.org/10.23919/SustainIT.2017.8379814 35. Ribeiro, M., Nisi, V., Prandi, C., Nunes, N.: A data visualization interactive exploration of human mobility data during the COVID-19 outbreak: a case study, pp. 1–6. IEEE (2020). https://doi.org/10.1109/ISCC50000.2020.9219552 36. Ribeiro, M., Nunes, N., Nisi, V., Sch¨ oning, J.: Passive Wi-Fi monitoring in the wild: a long-term study across multiple location typologies. Pers. Ubiquit. Comput., 1–15 (2020). https://doi.org/10.1007/s00779-020-01441-z 37. Ruiz-Ruiz, A.J., Blunck, H., Prentow, T.S., Stisen, A., Kjærgaard, M.B.: Analysis methods for extracting knowledge from large-scale WiFi monitoring to inform building facility planning. Presented at the (2014). https://doi.org/10.1109/ PerCom.2014.6813953 38. Said, M., Samuel, M., Shannan, N., Bashir, F.M., Dodo, Y.: Novel vision-based thermal people counting tool for tracking infected people with viruses like COVID19. J. Adv. Res. Dyn. Control Syst. 12, 1115–1119 (2020). https://doi.org/10.5373/ JARDCS/V12SP7/20202210 39. Shaw, P., Mikusz, M., Nurmi, P., Davies, N.: Tacita: a privacy preserving public display personalisation service. Presented at the (2018). https://doi.org/10.1145/ 3267305.3267627 40. Sohn, T., et al.: Mobility detection using everyday GSM traces. In: Dourish, P., Friday, A. (eds.) UbiComp 2006. LNCS, vol. 4206, pp. 212–224. Springer, Heidelberg (2006). https://doi.org/10.1007/11853565 13 41. Stevens, H., Haines, M.B.: TraceTogether: pandemic response, democracy, and technology. East Asian Sci. Technol. Soc. 14(3), 523–532 (2020). https://doi.org/ 10.1215/18752160-8698301 42. Tang, X., Xiao, B., Li, K.: Indoor crowd density estimation through mobile smartphone Wi-Fi probes. IEEE Trans. Syst. Man Cybern. Syst. 50(7), 2638–2649 (2020). https://doi.org/10.1109/TSMC.2018.2824903. Conference Name: IEEE Transactions on Systems, Man, and Cybernetics: Systems 43. Vedaei, S.S., et al.: COVID-SAFE: an IoT-based system for automated health monitoring and surveillance in post-pandemic life. IEEE Access 8, 188538–188551 (2020).https://doi.org/10.1109/ACCESS.2020.3030194. Conference Name: IEEE Access 44. Whitelaw, S., Mamas, M.A., Topol, E., Spall, H.G.C.V.: Applications of digital technology in COVID-19 pandemic planning and response. Lancet Digital Health 2(8), e435–e440 (2020). https://doi.org/10.1016/S2589-7500(20)30142-4 45. Whittle, J., et al.: VoiceYourView: collecting real-time feedback on the design of public spaces. Presented at the (2010). https://doi.org/10.1145/1864349.1864358

24

M. Ribeiro et al.

46. Wissel, B.D., et al.: An interactive online dashboard for tracking COVID-19 in U.S. counties, cities, and states in real time. J. Am. Med. Inform. Assoc. 27(7), 1121–1125 (2020). https://doi.org/10.1093/jamia/ocaa071 47. Zhao, X., Delleandrea, E., Chen, L.: A people counting system based on face detection and tracking in a video, pp. 67–72. IEEE (2009). https://doi.org/10.1109/ AVSS.2009.45

Hanging Out Online: Social Life During the Pandemic Ashwin Singh1(B) and Grace Eden2 1

2

Living Lab, IIIT Delhi, New Delhi, India [email protected] Department of Human-Centered Design, IIIT Delhi, New Delhi, India [email protected]

Abstract. In March 2020, the government of India ordered a nationwide lockdown to prevent the spread of Covid-19. This led to the shutdown of educational institutes throughout the country, restricting all activities to online mediums. The shift has affected how students engage with each other, where rather than in-person interaction, they meet through a variety of online tools. In this paper, we discuss how the normal everyday routine of ‘hanging out’ with friends has been transformed during a prolonged lockdown of over ten months and counting. We investigate the opportunities and challenges students encounter when socializing online through various online modes including video calls, communal movie watching and social media. We discuss how social interaction; in particular, hanging out with friends has been transformed through these technologies and its implications for facilitating spontaneous interaction, negotiating intimacy, mutual understanding, and accessibility to different social groups. Finally, we conclude with a discussion of how these factors impact the transition from in-person to online modes of casual social interaction. Keywords: Qualitative fieldwork online

1

· Covid-19 · Negotiating intimacy

Introduction

During the Covid-19 pandemic, countries imposed nationwide lockdowns as a preventive measure against the spread of the disease [1]. In India, the closing of educational institutes affected over 34 million university and college-going students [2]. In addition, 71% of youth located in metropolis cities were spending the lockdown with their families [3]. With all academic activities restricted to online mediums and restrictions on “hanging out” in physical settings, the social lives of university students have been significantly affected. Now, it is only possible to hang out with friends online. In this study, we present preliminary findings on how students maintain their social connections, what we refer to as “hanging out”; specifically how they make use of various technologies to interact with friends online while adapting to the c IFIP International Federation for Information Processing 2021  Published by Springer Nature Switzerland AG 2021 C. Ardito et al. (Eds.): INTERACT 2021, LNCS 12933, pp. 25–33, 2021. https://doi.org/10.1007/978-3-030-85616-8_2

26

A. Singh and G. Eden

lockdown. Our aim is to understand which technologies facilitate communication amongst the students, with a focus on the role technologies play in mediating everyday social interaction [4]. In addition, we investigate the different ways in which students use computer-mediated communication (CMC) technologies to share their experiences with friends and how they manage those relationships. We conduct a preliminary qualitative study to investigate the effects of the lockdown on the social life of university students; particularly the transformations brought about by CMC technologies and how they unfolded due to differences between in-person and computer-mediated interactions. We discuss these differences and their implications for opportunities for spontaneous social interaction. Next, we talk about the shifts in context and social accessibility of social interaction during the lockdown, followed by how students negotiate intimacy and configure their identity in CMC technologies. Finally, we conclude with a discussion of how students have adapted to the lockdown; discussing some challenges they encountered in the process.

2

Related Work

Our work builds upon three broad areas of computer-mediated communication, namely presence, presentation of self and common ground. Presence plays an important role in how students use different CMC technologies to socialize with their peers whereas self-presentation is key to how they construct their identity online. Lastly, we examine examples of how common ground is developed in the CMC technologies that students use and how these transform everyday social interactions. 2.1

Being Together Online

While presence is concerned with the sense of being transported to a space different from a physical environment, mediated presence refers to social interaction that takes place through the medium itself [5] and can be defined as the degree to which people overlook the mediated nature of an interaction [6]. For this paper, we discuss two aspects of mediated presence, namely co-presence and social presence. Co-presence refers to the sense of being together with other people [7] and is reflexive in nature [8]. Specifically, it relates to how people perceive others through the medium, as well as how they are perceived by others [8]. In computer-mediated interaction, the level of co-presence available plays an important role in determining the extent to which technologies provide people with a sense of being together. Social presence, on the other hand, is the moment-to-moment awareness of other people and participation with them [7]. Social presence influences the extent to which interpersonal relationships are facilitated by the medium [9]. Levels of social presence are said to increase when multiple modalities (such as audio and video), low latency (immediate feedback during an interaction), and personalisation are available [10]. In our research, we are interested to explore how support for social presence is used as a criteria when students select technologies for interacting with friends online.

Hanging Out Online: Social Life During the Pandemic

2.2

27

Configuring Identity

Goffman’s work on the presentation of self describes how people’s roles differ depending upon the social context and the norms within it [11]. For example, people may underplay or overplay aspects of themselves such as vulnerability or formality to ensure compatibility with the idealised version of themselves that they wish to display to others. Self-presentation becomes particularly relevant in the use of social media where people’s profiles are visible to their different social circles including friends, acquaintances, co-workers and family members [12]. Each of these groups rarely mingle together in the real world [13]. Thus, when these diverse groups are flattened into a shared online space, it results in a phenomenon referred to as context collapse [12]. This situation results in a limiting effect on how people present themselves to others. Instead, people adopt a plethora of measures to navigate these tensions, such as using multiple accounts or creating alternate accounts under pseudonyms or nicknames [14]. Recently, people have started using private Instagram accounts (known as Finstas) to create intimate spaces that are only accessible to a trusted group of friends. This creates an opportunity for people to create safe spaces where they can share alternate identities with smaller groups [15]. Having control over one’s presentation of self plays a crucial role in regulating one’s identity, allowing more room for disclosure and affording more agency [16]. 2.3

Shared Experiences

Common ground also plays an important role in the development of shared experiences. It refers to the shared knowledge, beliefs and suppositions shared in common between a group of people [17]. Here, we discuss two dimensions of common ground - namely cultural common ground and personal common ground paying particular attention to the different roles they play in sustaining social interaction online. When people have an affiliation to a community that shares the same history and practices, for instance, students who belong to the same educational institution; they share a cultural common ground because their joint (or shared) knowledge is a result of their affiliation to the same institution [17]. On the other hand, personal common ground is constituted by joint personal events or experiences, which can act as shared bases for holding conversations [17]. Friends can share a personal common ground because they share personal experiences in the past, such as attending the same class, travelling, or hanging out together. These often are a result of events and experiences that occur during a typical day. However, individuals can have varying levels of personal common ground with different groups of people; for instance, people share a more extensive common ground with partners through the sharing of more private information as compared to their friends [17]. Challenges to the development of personal common ground is highlighted during the Covid-19 pandemic due to the shift from in-person to online interactions and the differing affordances of both in supporting shared experiences. Cultural and personal common ground form important contexts in which everyday social interactions are situated and

28

A. Singh and G. Eden

are important for developing a trove of shared experiences that foster interpersonal relationships [17]. In what follows, we discuss how its absence during the lockdown has brought challenges to students’ casual social interactions online.

3

Methodology

To understand how hanging out with friends has transformed students’ casual social interactions during the lockdown, we conducted a preliminary qualitative study with five university students. Participants were recruited using convenience sampling [18] owing to the ease of accessibility during the lockdown period. Upon approval from our Institutional Review Board, we conducted semistructured interviews for between 45 min to 1 h with five participants (M = 3, F = 2) (Table 1). The aim of the interviews was to understand different student perspectives on how they foster their social relationships online - specifically their normal routine of just hanging out with each other. Table 1. Description of participants. Id Age Gender Alias P1 21

woman Neetika

P2 21

man

Ketan

P3 21

man

Rahul

P4 20

woman Preeti

P5 20

man

Madhur

During the interviews we asked participants to describe how they would hangout with their friends and acquaintances before the lockdown and how they hang out with them now, during the lockdown. We asked them to describe their use of different technologies - when they were used, who they were used with, and why they were used. Finally, we asked them to describe the challenges they encountered while socializing online and how they managed their relationships during the lockdown. Interviews were conducted remotely and audio recorded using an online meeting tool. These were transcribed and anonymized with participant IDs. To analyze the data, we adopted an inductive coding approach to thematically code [19] the transcripts. In the first iteration, the first author identified high-level codes and for the second iteration, both authors performed line-by-line coding of the first round of high level categories. These refined codes were compared and discussed amongst the researchers to identify patterns and to reach consensus during remote data analysis sessions where the researchers discussed emerging themes, patterns and their significance to each other.

Hanging Out Online: Social Life During the Pandemic

4

29

Findings

In this section, we present findings from the study in four subsections: First, we discuss how lack of presence has transformed spontaneous interaction over the lockdown period and how it affects the routine practice of hanging out with friends. Second, we examine contexts which give rise to social interaction and how technologies mediate them in a way that might compromise the continued development of common ground. Third, we discuss how students negotiate intimacy and their presentation of self to engage in joint personal experiences and manage their social relationships online. Finally, we summarize students’ challenges when transitioning to the lockdown, including how online social interaction helped them during the transition period in the face of various technical and personal challenges. 4.1

Spontaneity of Social Interaction

In this section, we discuss how differences emerging from the affordances of inperson and mediated interactions affected online social interaction for students; particularly how the absence of a shared physical space has implications for spontaneity and situatedness of social interaction. Participants associated their socializing experiences prior to the lockdown with “spontaneity” because they could simply “bump into each other” in shared physical settings. “We were all living in a common space with similar schedules so there were naturally interactions between us while going to classes” (Ketan). However, after the lockdown, the absence of shared physical spaces affected this spontaneity, making social interaction more planned. In this shift, co-presence is diminished; resulting in students no longer feeling that they’re together with their friends. Rather, now they needed to make an effort to initiate conversations and chit chat with their friends. “Now, you talk to people you explicitly want to talk to because that call button on your phone needs to be pressed. Spontaneity is good but it gets you in situations you don’t want to be” (Rahul). Here we see that the deliberate action required to communicate with others online has resulted in selective interactions with specific people, allowing for a greater sense of control in mediated settings as compared to in-person interactions. All participants reported that their pre-lockdown socializing experiences were predominantly situated in their college surroundings. “In college you would just sit and eat together and talk about nothing. Now that’s very rare, right?” (Neetika). During the lockdown, situatedness has shifted into virtual settings that mediate interaction between users. However, this shift has resulted in a decline in social interaction. “My friend mentioned the same concern that there has been a sort of disconnect between us due to a reduction in communication during the lockdown” (Ketan). The lack of shared physical space has reduced the frequency of casual everyday exchanges between people with the “disconnect” felt by most participants illustrating the inadequacies of computer-mediated communication when the goal of these interactions is aimed at fostering relationships rather than the less nuanced goal of exchanging information.

30

4.2

A. Singh and G. Eden

Shifts in Context and Social Accessibility

Before the lockdown the context which acts as the basis for people to socially interact with each other emerged from personal experiences that occurred in everyday situations, many of which were shared with other students from college. “So we used to talk about our workload, TV shows, plans or even gossip about what’s going around in college” (Preeti). However, the dissolution of the ongoing establishment of in-person context and joint personal experiences has led to interactions which arise from a “requirement” to talk to college acquaintances to complete coursework online. “I don’t really get time to interact with people I’m not close to, unless it’s something particular like coursework” (Rahul). Interactions motivated by academic work from college indicate a shift in context and show how cultural common ground sustains some level of social interaction between people who were not close to each other. However, due to the lack of personal experiences together, there is less opportunity for spontaneous conversations to develop. Another consequence of the shift to online mediums is seen in the different social groups people interact with both pre-lockdown and now. “I think, before the lockdown, I used to socialize with my school friends less. It’s definitely a lot more now since you reach everyone that way [online]” (Neetika). Amidst the absence of in-person interaction access to people from different social groups (e.g. school and college friends) in the online mode has now flattened. Due to the affordances of computer-mediated communication technologies access across each of them is now the same whereas pre-lockdown these groups were associated with specific locations and circumstances. One result of this has been that students began reaching out to their old friends from school rather than trying to making new friends with college acquaintances online; something that would have developed more easily in-person. 4.3

Negotiating Intimacy

Technologies can facilitate interactions between people who share different types of personal common ground and students make decisions about which ones to use in order to negotiate different levels of intimacy with friends and partners. We examine instances of this negotiation across three different technologies, namely communal watching, video conferencing, and social media. Finally, we discuss how students overcome some of these tensions through how they configure and reveal their different online identities with different people. Communal Watching: Students engaged in communal watching with their close friends and partners over popular technologies such as Netflix Party and streaming over Discord. While watching TV shows and movies together was framed as an intimate activity, the degree of intimacy is negotiated. “Yeah, I use it [Netflix Party] quite often with my really close friends who watch the same shows that I do and we like to talk over chat” (Neetika). “Sometimes me and my girlfriend video call while watching a movie but keeping each other on mute so we’re there

Hanging Out Online: Social Life During the Pandemic

31

together but not actually speaking to each other” (Rahul). Neetika and her close friends use the chat feature on Netflix Party to engage in small talk throughout the film. Whereas Rahul and his girlfriend share a video call over Google Duo as the film plays only occasionally unmuting themselves to talk about the plot. These differences in the use of a communication side medium are intertwined with participants’ mutually desired level of intimacy and associated social presence (e.g. small talk between friends or silent presence between partners). Video Conferencing: In addition to its use in communal TV and movie watching, video conferencing is also used by students to talk to their close friends and partners; both being groups with whom they share an extensive personal common ground. Video conferencing affords a high degree of social presence due to the richness of both video and audio modalities alongside their synchronous nature which affords immediate feedback. These applications were described as intimate mediums where they felt comfortable interacting with not more than “three to four” people or they interacted one-on-one with their partner. While birthdays are a common occasion where large groups of people participate in video calls, some students said that it is often awkward for them due to a large number of people on the call that they do not know - this limits their selfpresentation in a medium which affords a high degree of social presence. Finstas: Another instance of context collapse occurred on Instagram, where students attempted to circumvent it by creating private secondary accounts known as Finstas (Fake Instagram accounts) to share their daily experiences with close friends. Madhur described his motivations for creating a Finsta referring to his account as a “repository of memories” whereas Neetika said that she wanted to socialise with her close friends beyond just texting or sharing memes. Finstas helped students indulge in self-disclosure with people they shared a deep personal common ground with. “Sharing experiences personally helps me process them as well as make my peace with them and I think it’s very wonderful that I am able to do so” (Ketan). Being a member of the LGBTQIA+ community, Ketan used his Finsta as a rite of passage, creating artistic concepts using animation, visuals, and poetry to share his experiences with trusted friends who accept his identity in contrast to his family and the larger group of followers on his public Instagram account. Finstas make it possible to share personal experiences by giving students control over who has access to content so that they can express the most personal and intimate aspects of themselves. 4.4

Adjusting to Lockdown

At the start of the lockdown, there was a period of uncertainty where it was unclear how soon the restrictions on in-person interaction would be lifted. Meeting online continued month-to-month and with it the realization that “social distancing” would be with us for the near future. As a result, students began to adapt to the “new normal”. “So I got into the habit of calling people every two,

32

A. Singh and G. Eden

three weeks and actively texting my close friends to keep in touch with them. So that’s definitely something that developed over time and became stable by June [2020]” (Neetika). Students would engage in conversations with their friends more often for emotional support. A common routine involved “checking-up” regularly to ensure that friends were doing alright. They discussed a fear of growing apart and expressed a desire to preserve their relationships. “I do wish to keep in touch more frequently with my friends because our bonds are something that need to be kept alive despite the physical distances between us now” (Ketan). However, online social interaction was described as draining and difficult to sustain over long periods of time. It has also been challenging for friends to interact with each other due to varying schedules and different time zones. These challenges also included consequences for personal privacy when students spent all their time at home. “There is some noise in the background. People keep coming in and going out so it’s difficult to maintain my personal space” (Preeti). At home individual agency can be compromised, creating obstacles to freely interacting with friends which would otherwise be uninterrupted when meeting in-person. Further, students gender and sexual identity was threatened due to heteronormative beliefs held at home forcing them to conceal aspects of themselves to ensure their safety.

5

Discussion

In this work we investigated how social interaction has been transformed during the Covid-19 lockdown for university students. We found that current technologies lack support for spontaneous interaction because they cannot simulate a sense of being together in the same way as a physical space. This absence of shared physical space makes it difficult to feel emotionally connected leading to a change in students’ interaction between their different social groups that are now equally accessible online. In fact, we found that students negotiate intimacy online by selecting which technologies to use with different people based upon their relationships with them (e.g. video, chat). These kinds of negotiations also allowed students to manage their personal identity when posting content on private Finsta accounts. Additionally, regular interactions between friends are used to check-in on each others’ well-being. However, students still face challenges exercising agency and managing personal space because they spend more time at home where they can be easily distracted and where the boundaries between personal, social, and academic spaces have become increasingly indistinguishable. Added to this, queer students felt their identity was restricted when confined at home. Further, different schedules between geographically distributed friends made it difficult for them to meet. While the ongoing lockdown in India continues to transform how friends engage with each other our findings present opportunities to reconsider how technologies could be designed to accommodate ‘hanging out’. To this end, our future work will involve a series of co-design workshops with a larger group of students to sketch out ideas for how to better support spontaneity, intimacy, and shared experiences when hanging out with friends in casual social interactions online.

Hanging Out Online: Social Life During the Pandemic

33

Acknowledgments. Thanks and appreciation to our participants for contributing their valuable time. We also thank the Center for Design and New Media (A TCS Foundation Initiative supported by Tata Consultancy Services) at IIIT-Delhi for supporting this research.

References 1. Withnall, A.: India coronavirus: Modi announces 21-day nationwide lockdown, limiting movement of 1.4bn people. Independent (2020) 2. Education: from disruption to recovery. UNESCO, 2020. https://en.unesco.org/ covid19/educationresponse 3. Kaul, A., Chapman, T.: Life in lockdown: a survey of India’s urban youth. Observer Research Foundation, 2020. https://www.orfonline.org/wp-content/ uploads/2020/07/ORF-Monograph-Life-In-Lockdown1.pdf 4. December, J., Randall, N.: World Wide Web 1997 Unleashed, 2nd edn. Sams, Indianapolis (1996) 5. Lee, K.M.: Presence, explicated. Commun. Theor. 14(1), 27–50 (2004) 6. Lombard, M.: Direct responses to people on the screen: television and personal space. Commun. Res. 22(3), 288–324 (1995) 7. Skarbez, R., Brooks, F.P., Jr., Whitton, M.C.: A survey of presence and related concepts. ACM Comput. Surv. 50(6), 1–39 (2017) 8. Goffman, E.: Behavior in public places: notes on the social organization of gatherings, vol. 01 (1963) 9. Parker, E., Short, J., Williams, E., Christie, B.: The social psychology of telecommunication. In: Contemporary Sociology, vol. 7, no. 32 (1978) 10. Daft, R., Lengel, R.: Organizational information requirements, media richness and structural design. Manage. Sci. 32, 554–571 (1986) 11. Goffman, E.: The Presentation of Self in Everyday Life, vol. 01. Harmondsworth, London (1956) 12. Marwick, A., Boyd, D.: I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media Soc. 13(1), 114–133 (2011) 13. Bowers, K.: Situationism in psychology: an analysis and a critique. Psychol. Rev. 80, 307–336 (1973) 14. Marwick, A., Boyd, D.: Networked privacy: how teenagers negotiate context in social media. New Media Soc. 16, 1051–1067 (2014) 15. Xiao, S., Metaxa, D., Park, J., Karahalios, K., Salehi, N.: Random, messy, funny, raw: finstas as intimate reconfigurations of social media, pp. 1–13 (2020) 16. Boyd, D., Marwick, A.: Boyd, D., Marwick, A.E.: Social privacy in networked publics: Teens’ attitudes, practices, and strategies (September 22, 2011). A Decade in Internet Time: Symposium on the Dynamics of the Internet and Society, September 2011. 17. Clark, H.H.: Using Language. ‘Using’ Linguistic Books. Cambridge University Press, Cambridge (1996) 18. Bernard, H.: Research Methods in Anthropology: Qualitative and Quantitative Approaches, pp. 148–149 (2011) 19. Miles, M. B., Huberman, A. M.: Qualitative data analysis: An expanded sourcebook (2nd ed.). Thousand Oaks, CA: Sage (1994)

Investigating Italian Citizens’ Attitudes Towards Immuni, the Italian Contact Tracing App Cristina Bosco(B)

and Martina Cvajner

University of Trento, 38068 Rovereto, TN, Italy [email protected], [email protected]

Abstract. This research investigates Italian citizens’ attitudes towards the contact tracing app Immuni, promoted by the Italian government to track the spread of Covid19 and prevent possible future out-breaks. More specifically, this paper tries to uncover the factors that have motivated individuals to uptake the app or fail to do so. We have used a variety of qualitative methods: firstly, a virtual ethnography has been conducted, analyzing thematically 3013 tweets with the hashtag #Immuni. To further investigate the motivation behind either the adoption or the refusal of Immuni, twenty semi-structured interviews have been carried out with potential and actual users. A doctor and a health official have been also interviewed as secondary users. The results show that multiple factors shape users’ attitudes towards Immuni and influence their willingness to adopt it. The most relevant have to do with the perceived benefits, perceived barriers, perceived efficacy of the app and the cues to action. The usability of the overall app is also discussed with a focus on some of the key issues identified by the users. Keywords: User research research methods

1

· Contact tracing apps · Qualitative

Introduction

In June 2020, the Italian government announced the launch of Immuni, a contact tracing app (hereafter, CTA) designed to help preventing Covid19’s diffusion by tracking citizens’ social exchanges and notifying them of potentially infected contacts [6]. By August 2020, only 5 million users had downloaded Immuni, 13% of the target population (all of people living in Italy above 14 years old) [1]. A marketing campaign - “Scarica Immuni Week” (Let’s download Immuni) – was consequently launched (October 5–12, 2020). The campaign notwithstanding, in early February 2021, only 10 million Italians had downloaded the app. The effectiveness of CTAs is contingent upon their popularity. Previous studies c IFIP International Federation for Information Processing 2021  Published by Springer Nature Switzerland AG 2021 C. Ardito et al. (Eds.): INTERACT 2021, LNCS 12933, pp. 34–42, 2021. https://doi.org/10.1007/978-3-030-85616-8_3

Investigating Attitudes Towards Immuni

35

have found that the spread of the virus could be controlled digitally if 56% of the overall population (80% of smartphone users) adopted the app [8]. Ferretti and colleagues [7] have argued that if 60% of the population adopted a CTA, a significant portion of asymptomatic contagions could be prevented. Why then has Immuni failed to gain adequate acceptance? This paper - through the combined use of a virtual ethnography conducted on Twitter, and semi-structured interviews with primary and secondary users - explores user motivations driving the acceptance’ or the rejection - of Immuni.

2

Related Work

Being able to accurately and rapidly trace contacts is proven to be a key strategy in slowing down the spread of a virus. By identifying those in contact with an infected individual early enough, it becomes possible to quarantine them before they continue to spread the virus without overloading the national health system [5]. Digital contact tracing used together with manual tracing can fasten the process of tracing contacts and contribute to contain the spread of the virus. In addition, digital contact tracing presents some striking advantages compared to the alternative. In fact, while manual tracing requires specialized personnel and resources, on top of being extremely time consuming [9], digital contact, if well-implemented, can be cheaper and faster. Furthermore, it allows widening the range of people informed, indeed the contagion does not only occur among people who know each other but it might take place among strangers. Thus, with the employment of digital contact tracing, it becomes possible to reach and inform all those potential contacts that were nearly impossible to achieve by relying merely on memory. Yet, the success or the failure of this system highly depends on how many people decide to uptake it. Since most European countries have opted for introducing the contact tracing system on a voluntary basis, it becomes extremely important to explore and understand what motivates people to either download or not download a contact tracing app. There is, however, very little research carried out on users’ attitudes towards CTAs. To our knowledge, the only study which tackles this issue qualitatively has been conducted in UK by Williams et al. [10]. The researchers run multiple focus groups to explore British citizens’ attitudes towards the contact tracing app designed and developed by the National Health System (NHS). Several relevant themes emerged in their research, mainly participants expressed a lack of knowledge and appropriate information regarding contact tracing apps. In addition, they worried about their privacy and the use of their sensitive data and they showed concerns over the app’s uptake arguing that the percentage of uptakes of the app was a crucial factor in its success, thus if not enough people downloaded the app, then it would have not been fully effective. However, this study was conducted before an official app was released. The fact that the app did not exist but was only a proposal by the time the investigation was undertaken, makes it more difficult for the participants to have a

36

C. Bosco and M. Cvajner

clear idea of the actual app and its actual functioning. This research aimed to expand on this previous study by attempting to overcome this limitation. In fact, this study investigated users’ attitudes towards Immuni, which was released and available for download prior the beginning of the data collection. Therefore, the participants had a clear idea of the contact tracing app as well as had the choice to either download it or not. This made possible to expand the research from behavioral intentions to actual behavior. Furthermore, this research not only seeks to investigate users’ motivation for up-taking contact tracing app, but also aims to expand on previous studies by adding the analysis of secondary users of the contact tracing app. Indeed, in this scenario, the users are not only those citizens who make use of Immuni, but also the contact-tracing operators who deal with the Immuni’s health-related procedure and make the overall system function. They can be defined as secondary users, as they interact with primary users and interact with the system and they rely on primary users to obtain information to make the system function, following the definition provided by Alsos [2], and their contribution and insights on the procedure behind Immuni might be a key factor in the evaluation of the app itself and determining its success.

3

Immuni

Immuni is based on two different mechanisms: the generation of a random key of exposure and Bluetooth low energy (BLE). Once Immuni is installed on a device (Device1), it generates an exposure key that is temporary, random and changed daily. By using BLE, the phone transmits a signal that is generated by the random exposure key and which contains a rolling proximity identifier (ID of Device1). When another device with Immuni (Device2) receives this signal, both devices save the rolling proximity identifier locally, in their memory. If the user of Device 1 tests positive for Covid19, s/he can upload, with the external help of an operator, a random exposure key into Immuni’s server, which derives the rolling proximity identifiers broadcasted recently by Device 1. At this point, Device 2, which periodically compares all the random keys shared in the server with locally saved rolling proximity identifiers, will find a match with Device 1 and it will – if the length of the contact and the distance between the devices is above certain thresholds - inform the user of potential exposure to the virus. Once risky exposure has been detected, Immuni advises the user to consult a general practitioner. It is the GP that may prescribe a precautionary quarantine or a nasal swab. If the test for Covid-19 is positive, the operators of the National Health Service can, with the user consent, insert the user’s random temporary exposure key on Immuni’s server, thus informing the other users potentially exposed.

Investigating Attitudes Towards Immuni

4 4.1

37

Methods Data Collection

The data were collected using three different methodologies: – a virtual ethnography on tweets gathered from the launch of the app on 15th of June until the 12th of October, the end of the“Scarica Immuni Week”, – twenty interviews with primary users run from 21st of October until the 27th of October – two interviews with secondary users undertaken on the 5th of November The virtual ethnography data were gathered from June 15 (the launch of the app) to October 12 (the conclusion of the Scarica Immuni Week). The relevant tweets were scraped from Twitter using the website www.vicinitas.io All the tweets, posted within two days, having the #immuni specific hashtag were collected. The data set (N = 30,532) was subsequently cleaned, deleting all tweets not in Italian, duplicates, replies or retweets (around 25,500 of the tweets gathered). All surviving tweets were read to check their pertinence. The final data set consisted of 3,013 tweets. The authors followed Braun and Clarke’s guidelines [4]. A thematic analysis of the tweets revealed several themes: (a) perceived benefits, (b) perceived barriers, (c) perceived efficacy, (d) cues to action, and (e) overall usability. These themes were used as a model for the 22 semi-structured interviews with primary users, collected between October 21–27, 2020. Along with socio-demographics, participants were asked questions about the perceived benefit (What, do you think, has encouraged people to download Immuni?), the perceived barriers (What do you think has encouraged people not to download Immuni?), the perceived efficacy (Would you say that Immuni has an active role in the fight against Covid19?), cues to action (How did you hear about Immuni?), and overall usability (Have you ever encountered problem using Immuni?). Participants were recruited by using two methods. Ten were recruited by posting the following notice on Instagram and Facebook: (Do you want to participate in a very cool initiative?), while ten were recruited by randomly searching for last names on the website www.paginebianche.it, the Italian national phone-book posted online. Simultaneously, the authors decided to interview two secondary users. On November 5, a physician and an employee of an ASL (the local unit of the national health service), both involved with Immuni, were interviewed concerning two main issues: a) How do you think Immuni fits within the contact tracing procedure? b) What do you think of Immuni and its efficacy? The interviewed sample consisted of 11 men and 11 women. Almost half of the sample (N = 9) was in the 20–30 age cohort, while the remainder was evenly distributed. Among those 11 were Immuni users and 9 stated they did not download the app at the time of the interview. User evaluation of Immuni was defined as all factors (perceived benefits, perceived barriers and perceived efficacy) that have an impact on the overall evaluation of the Immuni as an effective CTA. Cues to action were defined as all the cues able to trigger the

38

C. Bosco and M. Cvajner

download of Immuni. Usability was employed as a general term covering all the usability aspects of the tool. All material gathered in this study was in Italian, yet it was translated to be presented in English in this paper. This generates a series of implications concerning the validity of the report and the research itself as it is almost impossible to achieve a perfectly matching translation as many words and idiomatic expressions in Italian might not be fully translatable in English. In this regard, translation has been done by the researcher, who is fully proficient in both English and Italian, with the goal of achieving semantic and conceptual equivalence [3]. 4.2

Data Analysis

Both datasets, consisting of tweets and interviews, were studied through inductive thematic analysis: a strict semantic logic, which exclusively paid attention to the meaning of words, was employed. All data were read to gain familiarity with the corpus and obtain a general understanding of it (phase 1). Subsequently, a systematic study of the textual corpus made identification of repeating patterns and the generation of codes possible. Themes and sub-themes emerged (phase 2). One of the researchers analyzed the data, while the second checked the results independently. To guarantee the anonymity of the tweets’ authors, all the collected data were anonymized. Interviewees were asked beforehand to give their informed consent and willingness to be recorded. They were informed that all data collected during the interview phase would be subsequently anonymized.

5

Results

5.1

Users’ Overall Evaluation of the App

The user’s evaluation is contingent upon three sub-themes: the perceived benefits arising from using the app, the perceived barriers encountered by the users when presented with Immuni, and the perceived efficacy of Immuni as a protective health measure. Concerning the benefits, the first sub-theme, the study documents how the benefits of Immuni are identified by the users mainly with tracing contacts effectively and contribution to prevent the spread of the virus. However, users show a variety of concerns regarding the use of Immuni, and these concerns acted as potential barriers that could prevent them from using it. The most significant one was the users’ suspect that Immuni could impose automatically a quarantine to all of those notified by the system, even when healthy. .. and above all, the fear of being in quarantine, without a real need to do so. (Giovanni)1 . The alleged quarantine imposed by Immuni was different from the one prescribed by their GP: it was not perceived as mandatory or as reliable as the one prescribed by a doctor, thus it could be easily avoided or dismissed by the users. And the fact that an app is the one which informs you that you have 1

interview(24/10/2020).

Investigating Attitudes Towards Immuni

39

to be in quarantine, and it is not your doctor, or a person whom you trust, or anybody else, it what convinces people not to download it (Lops)2 . Some users perceived this (alleged) imposition and the fact that Immuni could trace contacts with other users as a violation of their individual freedom and a violation of their privacy. This belief might have encouraged some not to trust the app and therefore, it might have led not to download it.#Immuni intrudes your privacy, it records your data and limits you freedom. Do not download it, for your own good and for those around you (Fla)3 . On the contrary, some participants showed no privacy fears or concerns claiming that they considered Immuni as a safe app and that they believed the data collected by Immuni to be safely protected. I mean, even if I have it on my phone, Immuni is an app that does not collect any information about me and I know that the data collected on Immuni are used only for epidemiological purpose. (Sara)4 . Another important aspect concerning the barriers encountered was the digital divide among users. This was mainly related to age and economic status. From the data analysis, it was highlighted how elderly, who are not used to digital devices, might encounter some difficulties in both downloading Immuni and using it correctly. In addition to that, Immuni, as it was described in the introduction, requires a newer software to function correctly. Thus, all of those citizens with older phones, who cannot afford a new phone to have Immuni, are not be able to download and use the app. Therefore, both elderly and people coming from a lower socio-economic status might be systematically excluded from this health-related public service. Often, the two factors (user’s older age and phone’s incompatibility) are combined together for the elderly, who have difficulties in using digital devices and smartphones, but also own older phones which might not be compatible with Immuni. Here you have some very old phones. For instance, my mum and dad, who are around 80s, don’t have smartphones, what will they do then? You can sit down and teach them how to use it but it is too complicated. Thus, Immuni has zero efficacy with such an old population and these very old phones (Salvatrice)5 . Concerning the last sub-theme: the evaluation of the efficacy of Immuni, it has emerged that Immuni was perceived as ineffective because of its low rate of adoption. According to many, such a failure had occurred because its adoption was completely voluntary, encouraging many citizens not to download it. Immuni was not imposed, thus it was not downloaded. (Moody)6 . It seems that many citizens would have preferred a more coercive method, wherein the government imposed Immuni to access some services and some public places They (the government) should have forced us to download it. We all should have been forced, also those who cannot do it by themselves, like me for example. They 2 3 4 5 6

interview(26/10/2020). tweet(10/10/2020). interview(23/10/2020). interview(22/10/2020). interview(21/10/2020).

40

C. Bosco and M. Cvajner

would have had to set up a public spot or office in which you can go and they help you out downloading it (Salvatrice)7 . Finally, concerning the evaluation of Immuni as an effective tracing system, the integration with the national health system (SSN) was discussed. According to the data analyzed, users believed that Immuni was not properly integrated into the actual operations of the national health system. Also because I know that there are some issues concerning the relation of Immuni with the ASL: I believe some operators don’t know how to insert the codes (Moody)8 . Such a suspicion appears unfounded, at least under the technological point of view. The data collected with the secondary users have made clear that doctors in charge of contact tracing had been carefully instructed on the use of Immuni and were all capable of making the overall Immuni’s system function correctly. Immuni is so easy, it takes only 3 min, it is not long at all. It is a very small part of the all procedure of contact tracing, not so significant concerning the time required (Contact tracing doctor)9 . 5.2

Cues to Action

The second main theme observed in the data gathered is the cues to action. Usually, they are either internal (physiological cues) or external (related to information transmitted by external factors such as friends, media, or institutions). This study has focused only on external cues. Almost all people interviewed expressed their disappointment towards the Immuni’s marketing campaign arguing that the app was not promoted enough and with enough insistence, and the marketing campaign, conducted by the government, did not neither reassure citizens’ regarding their privacy fear nor explain adequately the functioning of the app Minister @robersperanza, as you can see from this hundreds of comments, most of the population did not understand how Immuni works. Why there is no informative advertisements on TVs, newspapers?#COVID19italia @MinisteroSalute (Mec)10 . This might have contributed to both spread confusion regarding the app and also dismissing it as a not relevant tool I think that if the government actually cared about this project, they would have done more to promote it (Lops)11 . 5.3

Immuni’s Usability

This theme concerns with all the characteristics of the app that can influence its perceived usability and can impact on the perception of the app itself. Among usability design choices, the sub-theme of notifications has emerged. At the beginning, users complained about push notifications being too frequent and too 7 8 9 10 11

interview(22/10/2020). interview (21/10/2020). interview(5/11/2020). tweet(10/10/2020). interview(26/10/2020).

Investigating Attitudes Towards Immuni

41

disruptive. In an app designed only to inform about possible contagions, notifications should only be informative and should represent an infrequent, thus relevant, phenomenon. Subsequently, users started to express worries concerning lack of feedback. Notifications had become not only sporadic but also not very evident. One user claimed that he had been informed of a risky exposition only after he had entered the app. I saw the notification only as I entered the app, I didn’t receive any notification, I didn’t get a notification like when you receive a message. Nothing like that (Senator)12 . This could be extremely problematic: although the app itself asks the user to check one’s status every day, the user has nothing to do within the app, so s/he is not encouraged or even reminded of opening the app regularly And the app says “open Immuni once a day to check its status”, but if you open the app, there is nothing to do there, so why would I open it everyday? (Demetra)13 . Many have perceived Immuni as defined by a lack of feedback as it does not provide any external stimuli or outputs on whether the system is working appropriately.

6

Discussion

As countries are facing several difficulties in running a successful vaccination campaign and the number of infected citizens remains dangerously high, digital contact tracing could be now even more than in the past a fundamental tool in containing the virus. Investigating all the factors that influence users’ perception and attitudes towards the Italian CTA has proved to be an extremely complex and challenging task. As a matter of fact, several multi-layered factors are involved in influencing users’ views and opinions concerning Immuni. We found that the overall evaluation of the app is affected by three main clusters: the perceived benefits, the perceived barriers and the overall efficacy of the tool. These three factors are highly intertwined and related with one and other. These findings are in line with previous research, which have found the importance of privacy concerns and doubts over the mass uptake of then app as two relevant factors impacting CTA’ citizen’s evaluation [10]. An external aspect that might have influenced users’ perception of Immuni is the official marketing campaign as well as all the information circulating, both online and offline, regarding the app. Some have claimed that such a lousy campaign may have actually led to more reluctance and skepticism towards the app, as the communication failed in clarifying the purpose of the app and the way it worked. It also did not dispel the privacy fears of the citizens. Concerning the overall usability, most people presented no issues in using the app, they found it simple to understand and easy to use. The main problems raised by primary users and by the virtual ethnography concerned with the design of exposure notification. Users are asked to check routinely the status of the app, but as there is no reminder or notification sent to the user, why would 12 13

interview(25/10/2020). interview(21/10/2020).

42

C. Bosco and M. Cvajner

s/he check the app every day? This proves to be a major usability barrier, as it might prevent some users from discovering quickly of their risky exposition. In conclusion, our research can shed a light on citizens’ attitudes towards CTAs and open up new avenues of research in the area of attitudes towards CTAs.

References 1. Agi.it, R.: Download, positivi, notifiche: i primi 3 mesi di immuni, September 2020. https://www.agi.it/cronaca/news/2020-09-14/immuni-download-positivinotifiche-9663280/ 2. Alsos, O.A., Svanæs, D.: Designing for the secondary user experience. In: Campos, P., Graham, N., Jorge, J., Nunes, N., Palanque, P., Winckler, M. (eds.) INTERACT 2011. LNCS, vol. 6949, pp. 84–91. Springer, Heidelberg (2011). https://doi.org/10. 1007/978-3-642-23768-3 7 3. Behling, O., Law, K.S.: Translating questionnaires and other research instruments: problems and solutions. Sage 133 (2000) 4. Braun, V., Clarke, V.: Using thematic analysis in psychology. Qual. Res. Psychol. 3(2), 77–101 (2006) 5. Dar, A.B., Lone, A.H., Zahoor, S., Khan, A.A., Naaz, R.: Applicability of mobile contact tracing in fighting pandemic (COVID-19): issues, challenges and solutions. Comput. Sci. Rev. 38, 100307 (2020) 6. Falletti, E.: Privacy protection, big data gathering and public health issues: COVID-19 tracking app use in Italy. In: Big Data Gathering and Public Health Issues: COVID-19 Tracking App Use in Italy, 2 January 2021 7. Ferretti, L., et al.: Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science 368(6491) (2020) 8. Hinch, R., et al.: Effective configurations of a digital contact tracing app: a report to nhsx. en, April 2020. https://github.com/BDI-pathogens/covid-19 instant tracing/blob/master/Report 9. Verrall, A.: Rapid audit of contact tracing for COVID-19 in New Zealand, April 2020. https://apo.org.au/sites/default/files/resource-files/2020-04/aponid303350.pdf/ 10. Williams, S.N., Armitage, C.J., Tampe, T., Dienes, K.: Public attitudes towards COVID-19 contact tracing apps: a UK-based focus group study. Health Expectations 24(2), 377–385 (2020)

Social Companion Robots to Reduce Isolation: A Perception Change Due to COVID-19 Moojan Ghafurian1(B) , Colin Ellard2 , and Kerstin Dautenhahn1,3 1

Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada {moojan,kerstin.dautenhahn}@uwaterloo.ca 2 Department of Psychology, University of Waterloo, Waterloo, ON, Canada [email protected] 3 Department of Systems Design Engineering, University of Waterloo, Waterloo, ON, Canada

Abstract. Social isolation is one of the negative consequences of a pandemic like COVID-19. Social isolation and loneliness are not only experienced by older adults, but also by younger people who live alone and cannot communicate with others or get involved in social situations as they used to. In such situations, social companion robots might have the potential to reduce social isolation and increase well-being. However, society’s perception of social robots has not always been positive. In this paper, we conducted two online experiments with 102 and 132 participants during the self isolation periods of COVID-19 (May-June 2020 and January 2021), to study how COVID-19 has affected people’s perception of the benefits of a social robot. Our results showed that a change caused by COVID-19, as well as having an older relative who lived alone or at a care center during the pandemic significantly and positively affected people’s perception of social robots, as companions, and that the feeling of loneliness can drive the purchase of a social robot. The second study replicated the results of the first study. We also discuss the effects of Big 5 personality traits on the likelihood to purchase a social robot, as well as on participants’ general attitude towards COVID-19 and adapting to the pandemic.

1

Introduction

Social isolation is a common issue among older adults, especially those with disabilities, and can affect health and quality of life. Lack of social support can lead to loneliness, which is a significant public health issue [15] and is associated with multiple negative health outcomes, such as depression [2,5], anxiety [42], and an increase in cardiovascular risk [46]. While social isolation is more common among older adults, pandemics such as COVID-19 can lead to an increase in social isolation not only in older adults [3], but among adults of all ages [14,48]. Social c IFIP International Federation for Information Processing 2021  Published by Springer Nature Switzerland AG 2021 C. Ardito et al. (Eds.): INTERACT 2021, LNCS 12933, pp. 43–63, 2021. https://doi.org/10.1007/978-3-030-85616-8_4

44

M. Ghafurian et al.

isolation is an important consequence of COVID-19, along with many other negative consequences [14] such as an increase in family violence [47]. As concluded by Van Bavel et al. [49], action is required to mitigate these devastating effects. Since the outbreak of COVID-19, many researchers have emphasized different benefits of using robots and artificially intelligent systems on mitigating the associated risks. These applications included using robots for medical education [17], managing the spread of COVID-19 in public areas by reducing human contact, disinfecting areas, providing security [53,54], and helping children to stay connected with each other [43]. We believe that another important application area is to use social robots as companions during a pandemic such as COVID-19. Social robots have the potential to act as companions and reduce social isolation, and have already been used successfully in providing care and support in many domains, such as by improving older adults’ mood [50,51], reducing depression [1], and even reducing the need to use medication [45]. However, perception of social robots has not always been positive, and previously, prior to COVID-19, it has been reported that there is an increasing trend in people’s negative attitudes towards robots [16]. It might also be difficult to perceive their benefits accurately, especially by those who do not have any experience of, or direct knowledge of anyone who has experienced loneliness and social isolation. One of the common concerns raised by many researchers and participants alike is that social robots may replace human companions. While this is a valid concern, social robots can be highly beneficial in situations where such human interaction is in fact prohibited, for example during a pandemic such as COVID-19, or for older adults who might be socially isolated, e.g., due to medical conditions. In these situations the presence of social robots could be positive and effective in reducing their loneliness and improving society’s mental wellbeing. While companion robots can be beneficial to reduce loneliness, they can only succeed if society has a positive attitude towards them. Specifically, people need to perceive the benefits of social robots and be willing to adopt them. It has been previously shown that lower social support can increase older adults’ acceptance of social robots that are less intuitive to use [4]. But, a lower level of psychological well-being, such as emotional loneliness, life satisfaction, and depressive mood can reduce acceptance, depending on the robot and how intuitive it is to use [4]. We argue that the changes caused by COVID-19 has resulted in a higher level of attention to the consequences of social isolation, which could highlight the benefits of having a companion robot (for self or loved ones), and could change society’s perception of them. Therefore, we investigated how experience of social isolation due to COVID19 has affected the perception of benefits of companion robots, and whether loneliness can affect the tendency to purchase a social robot (as an indicator of participants’ intention to use it). We further asked what tasks people would prefer for a companion robot to carry out, and what aspects of a companion robot they perceived to be important.

Social Companion Robots to Reduce Isolation

45

Research Questions and Hypotheses: This paper investigated seven research questions: RQ1: Has a change due to COVID-19 affected people’s perception of social companion robots? RQ2: Has having an older relative who lives alone or at a care center during the pandemic affected people’s perception of social companion robots? RQ3: Can loneliness increase the likelihood of purchasing a companion robot? RQ4: Do people have strong preferences for tasks of companion robots? RQ5: Are social elements (e.g., showing emotions) perceived to be important for a companion robot, and how does the importance of such components compare with the technical accuracy of the robots? RQ6: How do different personality types affect attitudes towards COVID-19, as well as attitudes towards social robots during the pandemic? RQ7: Are the findings related to the perception of social robots due to COVID19 and the effect of loneliness on purchase of a social robot robust and independent of the time that the study was conducted? Our hypotheses were as follows. H1: COVID-19 has caused people to reflect more about the consequences of social isolation, and has positively affected attitudes towards social companion robots, because the pandemic pointed out situations where social robots can fill gaps in the provision of social interaction, as opposed to robots being perceived negatively as replacements of human contact. H2: For the same reason as in H1, having an older relative who either lives alone or at a care center has positively affected the attitude towards social robots. H3: A change in perception of social robots can positively influence people’s tendency to purchase one. H4: Loneliness can be an important factor in adopting a social robot and can increase the tendency to purchase a social companion robot. H5: Social elements such as robots’ ability to show emotions, adapting its behaviour based on its users, etc., will be perceived to be as important as technical accuracy by people, since many studies have shown their advantages in experimental settings. H6: The observed effects can be replicated at a later time, as we believe that these effects (e.g., change of perception of social robots due to COVID-19) are long-term, as a result of COVID-19, and independent of the time of the study.

2

Background

In previous studies, social robots were successfully used in many contexts, such as for increasing older adults’ social engagement [34,41] and providing companionship [26,32]. They also helped with activities of daily living, such as helping older adults with dementia with the process of eating food [13], as well as supporting

46

M. Ghafurian et al.

nurses in a hospital [37]. Furthermore, social robots have been able to help children with autism in many domains such as therapy and education [11,39], for example by increasing their social engagement [40], or even involving children in peer-learning scenarios, e.g., to improve children’s writing skills [10]. However, despite their success in multiple domains and a variety of studies, many of the existing social robots have not been used much beyond the context of these studies. A part of this can be due to the existing challenges, such as having robots that can act fully autonomously, or concerns related to a robot’s monetary cost [28,35] (as it is challenging to build affordable consumer robots [44]). To be successfully adopted by their intended users, multiple factors are critical, which can be related to the robot (e.g., its appearance and design, its capabilities, etc.) and its users (e.g., users’ acceptance, trust, attitude towards robots, and likelihood of adopting a robot). User acceptance is in fact essential for the success of social robots [44]. According to a recent study conducted in Europe [16], people have become more cautious in using robots, and their attitude towards autonomous robots has become more negative during the period 2012 to 2017. This study also showed that people are more comfortable with robots at workplaces, as opposed to the robots used in healthcare and those that are designed to help older adults. According to this study, participants’ age, gender, and education level, and employment status could all affect one’s attitude towards robots: men were found to have a more positive attitude towards robots and education was positively correlated with a positive attitude [16]. The growing anxiety in using robots was associated with the concern that the robots may replace humans and lead to losing one’s job [8,27,29]. A negative attitude towards social robots can not only affect their acceptance [38] and adoption (as people with a highly negative attitude tend to avoid human–robot communication [30]), but also can affect people’s interactions with the robot, and as shown by Nomura et al., users’ self-expression towards the robots [30]. In general, society’s perception and acceptance of social robots are affected by multiple factors. For example, de Graaf et al. recommended creating a clear purpose for robots, increasing robots’ social capabilities, and considering the use context (e.g., living situation, time and location of use) to be important for developing social robots that are accepted by society [19]. In the context of healthcare, many factors such as a robot’s design, personality, adaptability, humanness, size, and gender can affect how people react to it and affect users’ acceptance [7]. Furthermore, it has been shown that the attitude towards social robots can significantly affect their acceptance [31]. There are multiple other factors that have been shown to affect acceptance of social robots, including but not limited to, feelings of social presence [21], perceived enjoyment [21], familiarity with robots [6], robots being functionally relevant [12], emotional displays (e.g., positive versus negative) [6], robots’ ease of use [12], level of interaction between potential users and robots [33], the category label assigned to the robot (i.e., the way the robot is introduced, e.g., a “home appliance”) [23], and even age [38].

Social Companion Robots to Reduce Isolation

47

Furthermore, previous studies showed different results regarding the impact of loneliness on the acceptance of robots: loneliness has been shown to be correlated with a more positive attitude towards robots (by increasing the perceived social presence) [24] and a more negative attitude (by decreasing anthropomorphic tendencies) [25] towards social robots. However, in many cases, social robots have been promising in reducing loneliness and social isolation, especially in older adults [36].

3

Experiment

To address our research questions, two online studies were conducted in May 2020 and January 2021 in Canada, both during the self isolation period of COVID19. While both were conducted during the times that people have been self isolating, the time difference enables us to ensure that the findings are robust and independent of participants’ level of experience with COVID-19. 3.1

Methodology

A questionnaire was designed and administered online on Amazon Mechanical Turk. The questionnaire measured different aspects of people’s experience of COVID-19 and perception of robots, and was implemented in 5 sections as below: Section 1 - Loneliness: Feeling of loneliness as measured through the 8-item UCLA Loneliness Scale Questionnaire (ULS-8) [20] Section 2 - Big 5 Personality: Big 5 personality, i.e., personality measured through five different traits of Extroversion, Openness to experience, Conscientiousness, Extraversion, Agreeableness, and Emotionality (or Neuroticism) was measured using the TIPI Questionnaire [18] Section 3 - Extroversion: The specific extroversion dimension of Big 5 was assessed via 24 items related to Extroversion from the 120-item Big 5 questionnaire (IPIP-120 Personality Test) [22], to ensure consistency and accuracy in the results of the Big 5 personality test. Note that the results of this questionnaire were used only for ensuring the consistency between TIPI and IPIP-120 results, as we preferred to avoid using all items from IPIP-120 (to avoid participant fatigue). Section 4 - Perception of Social Robots: A questionnaire was designed, asking participants: (a) whether they were likely to purchase a social robot if it is “affordable” for them (rated on a continuous scale from “not at all” to “very likely”). (b) if their perception of social robots as companions has changed due to COVID-19 (rated on a continuous scale from “Not Changed at All” to “Completely Changed”), with a follow up question asking them to explain why it has/has not changed (answer provided as a text entry). (c) what task/tasks they preferred for a companion robot during COVID-19. The choices were: music/video, chitchat and joking, dancing, exercises, reminders (health related, e.g., medicine and appointments, or daily life/socially related),

48

M. Ghafurian et al.

cooking, games, relaxation or meditation or breathing exercises, storytelling or reading together, and Other (we asked them to specify). (d) what appearance they preferred a social robot to have. The choices were: Human-like, Animal-like, and Other (we asked them to specify). (e) what were the elements in a social robot that were important to them. The choices were: Not making mistakes, Ability to show emotions, Not requiring much maintenance, Having a specific behaviour that they might like, Recognizing them, and Other (please specify). All questions that were not rated on a continuous scale (i.e., preference about the robot’s appearance, important elements for a social robot, and preferred tasks) were provided as multiple-choice questions, where participants could select as many responses as they wished, with the response of “Other” having a text entry to be completed. Section 5 - Demographics: A questionnaire was designed to gather participants’ demographics information, as well as other information that could affect their attitude and behaviour during COVID-19, which included: (a) age, (b) gender, (c) if participants had pets/type of pets, (d) number of people in their household, (e) number of children in their household, (f) if they have a relative who is over 80 and lives alone (if yes, how can a social robot help them?), (g) if they have a relative who is over 80 and lives in a care home (if yes, how can a social robot help them?), (h) how can a social robot help them stay connected with an older relative, (i) how they feel about social isolation and COVID-19 (responses ranged from their life is completely and negatively affected to their life is completely and positively affected), and how they feel about the way their life has changed, (k) aside from work communications, how many people did they connect with and if this number was changed due to COVID-19, (l) whether they have ever moved to a new country and lived there for a significant period for a purpose other than a holiday (as an indication of their ability to adapt to significant changes in their lives), (m) how stressed/anxious they were due to COVID-19, (n) how much they followed social isolation rules, and (o) if they thought there were differences between social robots and conversational virtual agents, to explain which one they prefer, and to explain why.1 On each page of the questionnaire, attention and sanity checks were added. For example, we presented the same question with the opposite direction of the scale for the answers, or questions with clear answers, such as “How much do you think that drinking water is liquid” or “Do you agree that Tuesdays are considered weekends in North America?”. Finally, we intended for participants to complete the study on a positive note. Thus, with the purpose of trying to change participants’ mood after thinking about the questions mentioned above, we asked them about their favourite animal, favourite cartoon character, and favourite movie. These questions were included at the very end and served solely as an attempt to end the study on a positive note. 1

Note that only a subset of these results, which were appropriate for the scope of this paper are discussed due to page limits.

Social Companion Robots to Reduce Isolation

3.2

49

Procedure

Upon reading the information letter and signing the consent form, participants were directed to the questionnaire. They completed the 5 sections of the questionnaire mentioned above. Afterwards they received the completion code and instructions on how to submit the HIT2 . 3.3

Participants

The first study was conducted May-June 2020. 110 participants were recruited on Amazon Mechanical Turk. Participation was limited to Canada3 and those who had an approval rate over 97% (to minimize the risk of getting low quality responses) and had completed at least 100 HITs4 (to ensure familiarity with the interface). Data of five participants were removed as they failed the attention checks. The data of three participants’ were removed due to repeated participation. This finally resulted in 102 participants (41 Female, 61 Male; age: [18,66], mean: 34.6 yrs). The number of people in participants’ household ranged from 1 to 8, with one not indicating a number and reporting “more than 5”. The number of children in participants’ household ranged from 0 to 6, with one not indicating a number and reporting “more than 5”. The majority (68 participants) reported to have no children in their household. The second study was conducted in January 2021. By checking the MTurk IDs prior to participation, we ensured that those who participated in the first study will not participate again. Recruitment criteria on Amazon Mechanical Turk was the same as before: 97% approval rate based on at least 100 HITs. But as country was also limited as before, we were not able to recruit more than 70 participants for the second study (who did not participate in the first one); therefore, the criteria was changed to an approval rate of 95% based on at least 50 HITs. A total of 138 participants were recruited (60 female, 77 male, and 1 unknown). Five participants failed the attention checks and their data were removed from the study. One participant’s data was removed due to missing data. This left 132 participants (59 Female, 72 Male, and 1 unknown; age: [25,72], mean: 33.6 yrs). The reported number of people in participants’ household ranged from 1 to 26. The number of children in participants’ household ranged from 0 to 5. Full Ethics clearance was received from the University of Waterloo’s Research Ethics Committees prior to running the studies. Participants were notified about the nature of the questions under foreseeable risks in the consent form and were informed that the questions might make them think about different aspects of social isolation during COVID-19 and the feeling of loneliness. They were also 2 3 4

Human Intelligence Task. Participation was limited to Canada, where rules on social distancing were precisely defined and followed by the majority of people. A Human Intelligence Task (HIT) is a task on Mechanical Turk (MTurk), which is completed by the volunteers on MTurk.

M. Ghafurian et al. 1

Completely Changed p