International Conference on Innovative Computing and Communications: Proceedings of ICICC 2019, Volume 1 (Advances in Intelligent Systems and Computing, 1087) 981151285X, 9789811512858

Citation preview

Advances in Intelligent Systems and Computing 1087

Ashish Khanna · Deepak Gupta · Siddhartha Bhattacharyya · Vaclav Snasel · Jan Platos · Aboul Ella Hassanien Editors

International Conference on Innovative Computing and Communications Proceedings of ICICC 2019, Volume 1

Advances in Intelligent Systems and Computing Volume 1087

Series Editor Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland Advisory Editors Nikhil R. Pal, Indian Statistical Institute, Kolkata, India Rafael Bello Perez, Faculty of Mathematics, Physics and Computing, Universidad Central de Las Villas, Santa Clara, Cuba Emilio S. Corchado, University of Salamanca, Salamanca, Spain Hani Hagras, School of Computer Science and Electronic Engineering, University of Essex, Colchester, UK László T. Kóczy, Department of Automation, Széchenyi István University, Gyor, Hungary Vladik Kreinovich, Department of Computer Science, University of Texas at El Paso, El Paso, TX, USA Chin-Teng Lin, Department of Electrical Engineering, National Chiao Tung University, Hsinchu, Taiwan Jie Lu, Faculty of Engineering and Information Technology, University of Technology Sydney, Sydney, NSW, Australia Patricia Melin, Graduate Program of Computer Science, Tijuana Institute of Technology, Tijuana, Mexico Nadia Nedjah, Department of Electronics Engineering, University of Rio de Janeiro, Rio de Janeiro, Brazil Ngoc Thanh Nguyen , Faculty of Computer Science and Management, Wrocław University of Technology, Wrocław, Poland Jun Wang, Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, Shatin, Hong Kong

The series “Advances in Intelligent Systems and Computing” contains publications on theory, applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually all disciplines such as engineering, natural sciences, computer and information science, ICT, economics, business, e-commerce, environment, healthcare, life science are covered. The list of topics spans all the areas of modern intelligent systems and computing such as: computational intelligence, soft computing including neural networks, fuzzy systems, evolutionary computing and the fusion of these paradigms, social intelligence, ambient intelligence, computational neuroscience, artificial life, virtual worlds and society, cognitive science and systems, Perception and Vision, DNA and immune based systems, self-organizing and adaptive systems, e-Learning and teaching, human-centered and human-centric computing, recommender systems, intelligent control, robotics and mechatronics including human-machine teaming, knowledge-based paradigms, learning paradigms, machine ethics, intelligent data analysis, knowledge management, intelligent agents, intelligent decision making and support, intelligent network security, trust management, interactive entertainment, Web intelligence and multimedia. The publications within “Advances in Intelligent Systems and Computing” are primarily proceedings of important conferences, symposia and congresses. They cover significant recent developments in the field, both of a foundational and applicable character. An important characteristic feature of the series is the short publication time and world-wide distribution. This permits a rapid and broad dissemination of research results. ** Indexing: The books of this series are submitted to ISI Proceedings, EI-Compendex, DBLP, SCOPUS, Google Scholar and Springerlink **

More information about this series at http://www.springer.com/series/11156

Ashish Khanna Deepak Gupta Siddhartha Bhattacharyya Vaclav Snasel Jan Platos Aboul Ella Hassanien •

•

•

•

•

Editors

International Conference on Innovative Computing and Communications Proceedings of ICICC 2019, Volume 1

123

Editors Ashish Khanna Department of Computer Science and Engineering Maharaja Agrasen Institute of Technology New Delhi, Delhi, India Siddhartha Bhattacharyya Department of Computer Science and Engineering Christ University Bangalore, India Jan Platos Department of Computer Science VŠB—Technical University of Ostrava Ostrava, Czech Republic

Deepak Gupta Department of Computer Science and Engineering Maharaja Agrasen Institute of Technology New Delhi, Delhi, India Vaclav Snasel Department of Computer Science VŠB—Technical University of Ostrava Ostrava, Czech Republic Aboul Ella Hassanien Faculty of Computers and Information Cairo University Giza, Egypt

ISSN 2194-5357 ISSN 2194-5365 (electronic) Advances in Intelligent Systems and Computing ISBN 978-981-15-1285-8 ISBN 978-981-15-1286-5 (eBook) https://doi.org/10.1007/978-981-15-1286-5 © Springer Nature Singapore Pte Ltd. 2020 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 Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Dr. Ashish Khanna would like to dedicate this book to his mentors Dr. A. K. Singh and Dr. Abhishek Swaroop for their constant encouragement and guidance and his family members including his mother, wife and kids. He would also like to dedicate this work to his (late) father Sh. R. C. Khanna with folded hands for his constant blessings. Dr. Deepak Gupta would like to dedicate this book to his father Sh. R. K. Gupta, his mother Smt. Geeta Gupta, his mentors Dr. Anil Kumar Ahlawat and Dr. Arun Sharma for their constant encouragement, his family members including his wife, brothers, sisters, kids and his students close to his heart. Professor (Dr.) Siddhartha Bhattacharyya would like to dedicate this book to his father late Ajit Kumar Bhattacharyya, his mother late Hashi Bhattacharyya, his beloved wife Rashni and his colleagues Anirban, Hrishikesh, Indrajit, Abhijit, Biswanath and Hiranmoy, who have been beside him through thick and thin.

Professor (Dr.) Jan Platos would like to dedicate this book to his wife Daniela and his daughters Emma and Margaret. Professor. (Dr.) Aboul Ella Hassanien would like to dedicate this book to his beloved wife Azza Hassan El-Saman.

Organizing Committee

ICICC 2019 Steering Committee Members General Chairs Prof. Dr. Vaclav Snasel, VŠB—Technical University of Ostrava, Czech Republic. Prof. Dr. Siddhartha Bhattacharyya, Principal, RCC Institute of Information Technology, Kolkata. Honorary Chair Prof. Dr. Janusz Kacprzyk, FIEEE, Polish Academy of Sciences, Poland. Conference/Symposium Chair Prof. Dr. Maninder Kaur, Director, Guru Nanak Institute of Management, Delhi, India. Technical Program Chairs Dr. Pavel Kromer, VŠB—Technical University of Ostrava, Czech Republic. Dr. Jan Platos, VŠB—Technical University of Ostrava, Czech Republic. Prof. Dr. Joel J. P. C. Rodrigues, National Institute of Telecommunications (Inatel), Brazil. Prof. Dr. Aboul Ella Hassanien, Cairo University, Egypt. Dr. Victor Hugo C. de Albuquerque, Universidade de Fortaleza, Brazil. Conveners Dr. Ashish Khanna, Maharaja Agrasen Institute of Technology, India. Dr. Deepak Gupta, Maharaja Agrasen Institute of Technology, India.

vii

viii

Organizing Committee

Publicity Chairs Dr. Hussain Mahdi, National University of Malaysia. Dr. Rajdeep Chowdhury, Academician, Author and Editor, India. Prof. Dr. Med Salim Bouhlel, University of Sfax, Tunisia. Dr. Mohamed Elhoseny, Mansoura University, Egypt. Dr. Anand Nayyar, Duy Tan University, Vietnam. Dr. Andino Maseleno, STMIK Pringsewu, Lampung, Indonesia. Publication Chairs Dr. D. Jude Hemanth, Associate Professor, Karunya University, Coimbatore. Dr. Nilanjan Dey, Techno India College of Technology, Kolkata, India. Gulshan Shrivastava, National Institute of Technology, Patna, India. Co-conveners Dr. Avinash Sharma, Maharishi Markandeshwar University (Deemed to be University), India. P. S. Bedi, Guru Tegh Bahadur Institute of Technology, Delhi, India. Mr. Moolchand Sharma, Maharaja Agrasen Institute of Technology, India.

ICICC 2019 Advisory Committee Prof. Dr. Vincenzo Piuri, University of Milan, Italy. Prof. Dr. Valentina Emilia Balas, Aurel Vlaicu University of Arad, Romania. Prof. Dr. Marius Balas, Aurel Vlaicu University of Arad, Romania. Prof. Dr. Mohamed Salim Bouhlel, University of Sfax, Tunisia. Prof. Dr. Aboul Ella Hassanien, Cairo University, Egypt. Prof. Dr. Cenap Ozel, King Abdulaziz University, Saudi Arabia. Prof. Dr. Ashiq Anjum, University of Derby, Bristol, UK. Prof. Dr. Mischa Dohler, King’s College London, UK. Prof. Dr. Sanjeevikumar Padmanaban, University of Johannesburg, South Africa. Prof. Dr. Siddhartha Bhattacharyya, Principal, RCC Institute of Information Technology, Kolkata, India. Prof. Dr. David Camacho, Associate Professor, Universidad Autonoma de Madrid, Spain. Prof. Dr. Parmanand, Dean, Galgotias University, UP, India. Dr. Abu Yousuf, Assistant Professor, University Malaysia Pahang, Gambang, Malaysia. Prof. Dr. Salah-ddine, Krit University Ibn Zohr, Agadir, Morocco. Dr. Sanjay Kumar Biswash, Research Scientist, INFOCOMM Lab, Russia. Prof. Dr. Maryna Yena S., Senior Lecturer, Medical University of Kiev, Ukraine. Prof. Dr. Giorgos Karagiannidis, Aristotle University of Thessaloniki, Greece.

Organizing Committee

ix

Prof. Dr. Tanuja Srivastava, Department of Mathematics, IIT Roorkee. Dr. D Jude Hemanth, Associate Professor, Karunya University, Coimbatore. Prof. Dr. Tiziana Catarci, Sapienza University of Rome, Italy. Prof. Dr. Salvatore Gaglio, University degli Studi di Palermo, Italy. Prof. Dr. Bozidar Klicek, University of Zagreb, Croatia. Dr. Marcin Paprzycki, Associate Professor, Polish Academy of Sciences, Poland. Prof. Dr. A. K. Singh, NIT Kurukshetra, India. Prof. Dr. Anil Kumar Ahlawat, KIET Group of Institutions, India. Prof. Dr. Chang-Shing Lee, National University of Tainan, Taiwan. Dr. Paolo Bellavista, Associate Professor, Alma Mater Studiorum - Università di Bologna. Prof. Dr. Sanjay Misra, Covenant University, Nigeria. Prof. Dr. Benatiallah Ali, Associate Professor, University of Adrar, Algeria. Prof. Dr. Suresh Chandra Satapathy, PVPSIT, Vijayawada, India. Prof. Dr. Marylene Saldon-Eder, Mindanao University of Science and Technology. Prof. Dr. Özlem ONAY, Anadolu University, Eskisehir, Turkey. Prof. Dr. Kei Eguchi, Department of Information Electronics, Fukuoka Institute of Technology. Prof. Dr. Zoltan Horvath, Professor, Kasetsart University. Dr. AKM Matiul Alam, Canada. Prof. Dr. Joong Hoon Jay Kim, Korea University. Prof. Dr. Sheng-Lung Peng, National Dong Hwa University, Taiwan. Dr. Daniela Lopez De Luise, CI2S Lab, Argentina. Dr. Dac-Nhuong Le, Hai Phong University, Vietnam. Dr. Dusanka Boskovic, University of Sarajevo, Sarajevo. Dr. Periklis Chat Zimisios, Alexander TEI of Thessaloniki, Greece. Dr. Nhu Gia Nguyen, Duy Tan University, Vietnam. Prof. Dr. Huynh Thanh Binh, Hanoi University of Science and Technology, Vietnam. Dr. Ahmed Faheem Zobaa, Brunel University London. Dr. Kirti Tyagi, Inha University in Tashkent. Prof. Dr. Ladjel Bellatreche, University of Poitiers, France. Prof. Dr. Victor C. M. Leung, The University of British Columbia, Canada. Prof. Dr. Huseyin Irmak, Cankiri Karatekin University, Turkey. Dr. Alex Norta, Associate Professor, Tallinn University of Technology, Estonia. Prof. Dr. Amit Prakash Singh, GGSIPU, Delhi, India. Prof. Dr. Abhishek Swaroop, Bhagwan Parshuram Institute of Technology, Delhi. Prof Christos Douligeris, University of Piraeus, Greece. Dr. Brett Edward Trusko, President and CEO (IAOIP) and Assistant Professor, Texas A&M University, Texas. Prof. Dr. R. K. Datta, Director, MERIT. Prof. Dr. Joel J. P. C. Rodrigues, National Institute of Telecommunications (Inatel), Brazil; Instituto de Telecomunicações, Portugal. Prof. Dr. Victor Hugo C. de Albuquerque, University of Fortaleza (UNIFOR), Brazil.

x

Organizing Committee

Dr. Atta ur Rehman Khan, King Saud University, Riyadh. Dr. João Manuel R. S. Tavares, Professor, Associado com Agregação FEUP— DEMec. Prof. Dr. Ravish Saggar, Director, Banarsidas Chandiwala Institute of Information Technology, Delhi. Prof. Dr. Ku Ruhana Ku Mahamud, School of Computing, College of Arts and Sciences, Universiti Utara Malaysia, Malaysia. Prof. Ghasem D. Najafpour, Babol Noshirvani University of Technology, Iran. Prof. Dr. Sanjeevikumar Padmanaban, University of Aalborg, Denmark. Prof. Dr. Frede Blaabjerg, President (IEEE Power Electronics Society), University of Aalborg, Denmark. Prof. Dr. Jens Bo Holm Nielson, Aalborg University, Denmark. Prof. Dr. Venkatadri Marriboyina, Amity University, Gwalior, India. Dr. Pradeep Malik, Vignana Bharathi Institute of Technology (VBIT), Hyderabad, India. Dr. Abu Yousuf, Assistant Professor, University Malaysia Pahang, Gambang, Malaysia. Dr. Ahmed A. Elngar, Assistant Professor, Faculty of Computers and Information, Beni Suef University, Beni Suef, Salah Salem Str., 62511, Egypt. Prof. Dr. Dijana Oreski, Faculty of Organization and Informatics, University of Zagreb, Varazdin, Croatia. Prof. Dr. Dhananjay Kalbande, Professor and Head, Sardar Patel Institute of Technology, Mumbai, India. Prof. Dr. Avinash Sharma, Maharishi Markandeshwar Engineering College, MMDU Campus, India. Dr. Sahil Verma, Lovely Professional University, Phagwara, India. Dr. Kavita, Lovely Professional University, Phagwara, India. Prof. Prasad K. Bhaskaran, Professor and Head of Department, Ocean Engineering and Naval Architecture, IIT Kharagpur.

Preface

We hereby are delighted to announce that VŠB—Technical University of Ostrava, Czech Republic, Europe, has hosted the eagerly awaited and much-coveted International Conference on Innovative Computing and Communication (ICICC 2019). The second version of the conference was able to attract a diverse range of engineering practitioners, academicians, scholars and industry delegates, with the reception of abstracts from more than 2200 authors from different parts of the world. The committee of professionals dedicated toward the conference is striving to achieve a high-quality technical program with tracks on innovative computing, innovative communication network and security and Internet of things. All the tracks chosen in the conference are interrelated and are very famous among the present-day research community. Therefore, a lot of research is happening in the above-mentioned tracks and their related sub-areas. As the name of the conference starts with the word ‘innovation,’ it has targeted out-of-box ideas, methodologies, applications, expositions, surveys and presentations helping to upgrade the current status of research. More than 550 full-length papers have been received, among which the contributions are focused on theoretical, computer simulation-based research and laboratory-scale experiments. Among these manuscripts, 129 papers have been included in the Springer proceedings after a thorough two-stage review and editing process. All the manuscripts submitted to ICICC 2019 were peer-reviewed by at least two independent reviewers, who were provided with a detailed review pro forma. The comments from the reviewers were communicated to the authors, who incorporated the suggestions in their revised manuscripts. The recommendations from two reviewers were taken into consideration while selecting a manuscript for inclusion in the proceedings. The exhaustiveness of the review process is evident, given the large number of articles received addressing a wide range of research areas. The stringent review process ensured that each published manuscript met the rigorous academic and scientific standards. It is an exalting experience to finally see these elite contributions materialize into two book volumes as ICICC 2019 proceedings by Springer entitled International Conference on

xi

xii

Preface

Innovative Computing and Communications. The articles are organized into two volumes in some broad categories covering subject matters on machine learning, data mining, big data, networks, soft computing and cloud computing, although given the diverse areas of research reported it might not have been always possible. ICICC 2019 invited five keynote speakers, who are eminent researchers in the field of computer science and engineering, from different parts of the world. In addition to the plenary sessions on each day of the conference, five concurrent technical sessions are held every day to assure the oral presentation of around 129 accepted papers. Keynote speakers and session chair(s) for each of the concurrent sessions have been leading researchers from the thematic area of the session. A technical exhibition is held during all the two days of the conference, which has put on display the latest technologies, expositions, ideas and presentations. The delegates were provided with a book of extended abstracts to quickly browse through the contents, participate in the presentations and provide access to a broad audience of the audience. The research part of the conference was organized in a total of 35 special sessions. These special sessions provided the opportunity for researchers conducting research in specific areas to present their results in a more focused environment. An international conference of such magnitude and release of the ICICC 2019 proceedings by Springer has been the remarkable outcome of the untiring efforts of the entire organizing team. The success of an event undoubtedly involves the painstaking efforts of several contributors at different stages, dictated by their devotion and sincerity. Fortunately, since the beginning of its journey, ICICC 2019 has received support and contributions from every corner. We thank them all who have wished the best for ICICC 2019 and contributed by any means toward its success. The edited proceedings volumes by Springer would not have been possible without the perseverance of all the steering, advisory and technical program committee members. All the contributing authors owe thanks from the organizers of ICICC 2019 for their interest and exceptional articles. We would also like to thank the authors of the papers for adhering to the time schedule and for incorporating the review comments. We wish to extend our heartfelt acknowledgment to the authors, peer reviewers, committee members and production staff whose diligent work put shape to the ICICC 2019 proceedings. We especially want to thank our dedicated team of peer reviewers who volunteered for the arduous and tedious step of quality checking and critique on the submitted manuscripts. We wish to thank our faculty colleagues Mr. Moolchand Sharma and Ms. Prerna Sharma for extending their enormous assistance during the conference. The time spent by them and the midnight oil burnt are greatly appreciated, for which we will ever remain indebted. The management, faculties and administrative and support staff of the college have always been extending their services whenever needed, for which we remain thankful to them.

Preface

xiii

Lastly, we would like to thank Springer for accepting our proposal for publishing the ICICC 2019 proceedings. Help received from Mr. Aninda Bose, Senior Editor, Acquisition, in the process has been very useful. New Delhi, India

Ashish Khanna Deepak Gupta Organizers, ICICC 2019

About This Book

International Conference on Innovative Computing and Communications (ICICC 2019) was held on 21–22 March at VŠB—Technical University of Ostrava, Czech Republic, Europe. This conference was able to attract a diverse range of engineering practitioners, academicians, scholars and industry delegates, with the reception of papers from more than 2200 authors from different parts of the world. Only 129 papers have been accepted and registered with an acceptance ratio of 23% to be published in two volumes of prestigious Springer Advances in Intelligent Systems and Computing (AISC) Series. Volume 1 includes the accepted papers of machine learning, data mining and big data tracks. This volume includes a total of 77 papers from these two tracks.

xv

Contents

Improving the Accuracy of Collaborative Filtering-Based Recommendations by Considering the Temporal Variance of Top-N Neighbors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pradeep Kumar Singh, Showmik Setta, Pijush Kanti Dutta Pramanik and Prasenjit Choudhury

1

Exploring the Effect of Tasks Difficulty on Usability Scores of Academic Websites Computed Using SUS . . . . . . . . . . . . . . . . . . . . . Kalpna Sagar and Anju Saha

11

Prediction and Estimation of Dominant Factors Contributing to Lesion Malignancy Using Neural Network . . . . . . . . . . . . . . . . . . . . . Kumud Tiwari, Sachin Kumar and R. K. Tiwari

21

Prediction of Tuberculosis Using Supervised Learning Techniques Under Pakistani Patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Muhammad Ali and Waqas Arshad

33

Automatic Retail Invoicing and Recommendations . . . . . . . . . . . . . . . . Neeraj Garg and S. K. Dhurandher

41

Modeling Open Data Usage: Decision Tree Approach . . . . . . . . . . . . . . Barbara Šlibar

57

Technology-Driven Smart Support System for Tourist Destination Management Organizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Leo Mrsic, Gorazd Surla and Mislav Balkovic

65

Ortho-Expert: A Fuzzy Rule-Based Medical Expert System for Diagnosing Inflammatory Diseases of the Knee . . . . . . . . . . . . . . . . Anshu Vashisth, Gagandeep Kaur and Aditya Bakshi

77

GA with k-Medoid Approach for Optimal Seed Selection to Maximize Social Influence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sakshi Agarwal and Shikha Mehta

97

xvii

xviii

Contents

Sentiment Analysis on Kerala Floods . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Anmol Dudani, V. Srividya, B. Sneha and B. K. Tripathy Recommendation System Using Community Identification . . . . . . . . . . . 125 Suman Venkata Sai Voggu, Yuvraj Singh Champawat, Swaraj Kothari and B. K. Tripathy Comparison of Deep Learning and Random Forest for Rumor Identification in Social Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 T. Manjunath Kumar, R. Murugeswari, D. Devaraj and J. Hemalatha Ontological Approach to Analyze Traveler’s Interest Towards Adventures Club . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Toseef Aslam, Maria Latif, Palwashay Sehar, Hira Shaheen, Tehseen Kousar and Nazifa Nazir Performance Analysis of Off-Line Signature Verification . . . . . . . . . . . . 161 Sanjeev Rana, Avinash Sharma and Kamlesh Kumari Fibroid Detection in Ultrasound Uterus Images Using Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 K. T. Dilna and D. Jude Hemanth Progressive Generative Adversarial Binary Networks for Music Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 Manan Oza, Himanshu Vaghela and Kriti Srivastava Machine Learning Approach for Diagnosis of Autism Spectrum Disorders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 Sai Yerramreddy, Samriddha Basu, Ananya D. Ojha and Dhananjay Kalbande Methodologies for Epilepsy Detection: Survey and Review . . . . . . . . . . 207 Ananya D. Ojha, Ananya Navelkar, Madhura Gore and Dhananjay Kalbande Scene Understanding Using Deep Neural Networks—Objects, Actions, and Events: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Ranjini Surendran and D. Jude Hemanth Scene Text Recognition: A Preliminary Investigation on Various Techniques and Implementation Using Deep Learning Classifiers . . . . . 233 N. Bhavesh Shri Kumar, Dasi Naga Brahma Krishna Sumanth Reddy, K. Sairam and J. Naren Computer-Aided Diagnosis System for Investigation and Detection of Epilepsy Using Machine Learning Techniques . . . . . . . . . . . . . . . . . . 243 J. Naren, A. B. Sarada Pyngas and S. Subhiksha

Contents

xix

Moments-Based Feature Vector Extraction for Iris Recognition . . . . . . 255 J. Jenkin Winston and D. Jude Hemanth A Novel Approach to Improve Website Ranking Using Digital Marketing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 Khyati Verma, Sanjay Kumar Malik and Ashish Khanna A Comparative Study on Different Skull Stripping Techniques from Brain Magnetic Resonance Imaging . . . . . . . . . . . . . . . . . . . . . . . 279 Ruhul Amin Hazarika, Khrawnam Kharkongor, Sugata Sanyal and Arnab Kumar Maji Predicting Academic Performance of International Students Using Machine Learning Techniques and Human Interpretable Explanations Using LIME—Case Study of an Indian University . . . . . . 289 Pawan Kumar and Manmohan Sharma Improved Feature Matching Approach for Detecting Copy-Move Forgery and Localization of Digital Images . . . . . . . . . . . . . . . . . . . . . . 305 Vanita Mane and Subhash Shinde Sentiment Analysis Using Gini Index Feature Selection, N-Gram and Ensemble Learners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315 Furqan Iqbal Text Summarization by Hybridization of Hypergraphs and Hill Climbing Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327 Hemamalini Siranjeevi, Swaminathan Venkatraman and Kannan Krithivasan Comparing Machine Learning Algorithms to Predict Diabetes in Women and Visualize Factors Affecting It the Most—A Step Toward Better Health Care for Women . . . . . . . . . . . . . . . . . . . . . . . . . 339 Arushi Agarwal and Ankur Saxena Evolution of mHealth Eco-System: A Step Towards Personalized Medicine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 Mohit Saxena and Ankur Saxena Genetic Variance Study in Human on the Basis of Skin/Eye/Hair Pigmentation Using Apache Spark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371 Ankur Saxena, Shivani Chandra, Alka Grover, Lakshay Anand and Shalini Jauhari An Improved Technique on Existing Neck and Head Support Systems for Cervical Dystonia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381 Rahul Dubey, Rahul Vishwakarma and Ashish Mishra

xx

Contents

Automatic Text Summarization Using Fuzzy Extraction . . . . . . . . . . . . 395 Bharti Sharma, Nitika Katyal, Vishant Kumar, Shivani and Amit Lathwal A Comparison of Machine Learning Approaches for Classifying Flood-Hit Areas in Aerial Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407 J. Akshya and P. L. K. Priyadarsini Cognitive Services Applied as Student Support Service Chatbot for Educational Institution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417 Leo Mrsic, Tomislav Mesic and Mislav Balkovic Feature Extraction and Detection of Obstructive Sleep Apnea from Raw EEG Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425 Ch. Usha Kumari, Padmavathi Kora, K. Meenakshi, K. Swaraja, T. Padma, Asisa Kumar Panigrahy and N. Arun Vignesh AIRUYA-A Personal Shopping Assistant . . . . . . . . . . . . . . . . . . . . . . . . 435 Yashi Rai, Aishwarya Raj, Kumari Suruchi Sah and Akash Sinha Adapting Machine Learning Techniques for Credit Card Fraud Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443 Bright Keswani, Prity Vijay, Narayan Nayak, Poonam Keswani, Saumyaranjan Dash, Laxman Sahoo, Tarini Ch. Mishra and Ambarish G. Mohapatra Ensemble Feature Selection Method Based on Recently Developed Nature-Inspired Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457 Jatin Arora, Utkarsh Agrawal, Prayag Tiwari, Deepak Gupta and Ashish Khanna Diagnosis of Parkinson’s Disease Using a Neural Network Based on QPSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471 Srishti Sahni, Vaibhav Aggarwal, Ashish Khanna, Deepak Gupta and Siddhartha Bhattacharyya Transfer Learning Model for Detecting Early Stage of Prurigo Nodularis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483 Dhananjay Kalbande, Rithvika Iyer, Tejas Chheda, Uday Khopkar and Avinash Sharma Optimization of External Stimulus Features for Hybrid Visual Brain–Computer Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Deepak Kapgate, Dhananjay Kalbande and Urmila Shrawankar Predicting the Outcome of an Election Results Using Sentiment Analysis of Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 Ankur Saxena, Neeraj Kushik, Ankur Chaurasia and Nidhi Kaushik

Contents

xxi

A Wide ResNet-Based Approach for Age and Gender Estimation in Face Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517 Rajdeep Debgupta, Bidyut B. Chaudhuri and B. K. Tripathy Bio-inspired Algorithms for Diagnosis of Heart Disease . . . . . . . . . . . . . 531 Moolchand Sharma, Ananya Bansal, Shubbham Gupta, Chirag Asija and Suman Deswal Emotion Detection Through EEG Signals Using FFT and Machine Learning Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543 Anvita Saxena, Kaustubh Tripathi, Ashish Khanna, Deepak Gupta and Shirsh Sundaram Text Summarization with Different Encoders for Pointer Generator Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551 Minakshi Tomer and Manoj Kumar Performance Evaluation of Meta-Heuristic Algorithms in Social Media Using Twitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559 P. Silambarasi and Kiran L. N. Eranki Heuristic Coordination for Multi-agent Motion Planning . . . . . . . . . . . . 569 Buddhadeb Pradhan, Nirmal Baran Hui and Diptendu Sinha Roy Optimization of Click-Through Rate Prediction of an Advertisement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579 N. Madhu Sudana Rao, Kiran L. N. Eranki, D. L. Harika, H. Kavya Sree, M. M. Sai Prudhvi and M. Rajasekar Reddy Prediction of Cervical Cancer Using Chicken Swarm Optimization . . . . 591 Ayush Kumar Tripathi, Priyam Garg, Alok Tripathy, Navender Vats, Deepak Gupta and Ashish Khanna Describing Image Using Neural Networks . . . . . . . . . . . . . . . . . . . . . . . 605 Atul Kumar, Ratnesh Kumar and Shailesh Kumar Shrivastava An Efficient Expert System for Proactive Fire Detection . . . . . . . . . . . . 613 Venus Singla and Harkiran Kaur Computational Intelligence for Technology and Services Computing Analysis of Sentiment Analysis Techniques . . . . . . . . . . . . . . . . . . . . . . 623 Puja Bharti, Amit Kant Verma, Ravi Raj and Gopal Krishna Detection of Parkinson’s Disease Using Machine Learning Techniques for Voice and Handwriting Features . . . . . . . . . . . . . . . . . . . . . . . . . . . 631 Nikita Goel, Ashish Khanna, Deepak Gupta and Naman Gupta

xxii

Contents

Audio–Video Aid Generator for Multisensory Learning . . . . . . . . . . . . 645 Reshabh Kumar Sharma, Aman Alam Bora, Sachin Bhaskar and Prabhat Kumar E-Labharthi—Information Management for Sustainable Rural Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655 Amar Nath Pandey, Shakti Pandey and Sangeeta Sinha Prediction of Celiac Disease Using Machine-Learning Techniques . . . . . 663 Agrima Mehandiratta, Neha Vij, Ashish Khanna, Pooja Gupta, Deepak Gupta and Ayush Kumar Gupta Detection of Devanagari Text from Wild Images Through Image Processing Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675 Manas Bhardwaj, Savitoj Singh, Ashish Khanna, Ankita Gupta, Deepak Gupta and Naman Gupta Dynamic Web with Automatic Code Generation Using Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687 Prerna Sharma, Vikas Chaudhary, Nakul Malhotra, Nikita Gupta and Mohit Mittal Detection of Garbage Disposal from a Mobile Vehicle Using Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 699 Shubhi Jain, Naman Gupta, Ashish Khanna, Ankita Gupta and Deepak Gupta Using Neural Network to Identify Forgery in Offline Signatures . . . . . . 711 Piyush Agrawal, Rahul Bhalsodia and Yash Garg Analyzing the Impact of Age and Gender on User Interaction in Gaming Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 721 Abid Jamil, Ch. M. Nadeem Faisal, Muhammad Asif Habib, Sohail Jabbar and Haseeb Ahmad Analysis of Prediction Techniques for Temporal Data Based on Nonlinear Regression Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731 Pinki Sagar, Prinima Gupta and Indu Kashyap Robust Denoising Technique for Ultrasound Images Based on Weighted Nuclear Norm Minimization . . . . . . . . . . . . . . . . . . . . . . . 741 Shaik Mahaboob Basha and B. C. Jinaga Comparison of Machine Learning Models for Airfoil Sound Pressure Prediction and Denoising for Airbots . . . . . . . . . . . . . . . . . . . . . . . . . . . 751 R. Kavitha and C. R. Srivatsan

Contents

xxiii

Local Texture Features for Content-Based Image Retrieval of Interstitial Lung Disease Patterns on HRCT Lung Images . . . . . . . . 761 Jatindra Kumar Dash, Manisha Patro, Snehasish Majhi, Gandham Girish and P. Nancy Anurag Multilevel Quantum Sperm Whale Metaheuristic for Gray-Level Image Thresholding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 775 Siddhartha Bhattacharyya, Sandip Dey, Jan Platos, Vaclav Snasel and Tulika Dutta Malaria Detection on Giemsa-Stained Blood Smears Using Deep Learning and Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789 Nobel Dang, Varun Saraf, Ashish Khanna, Deepak Gupta and Tariq Hussain Sheikh An Effective Instruction Execution and Processing Model in Multiuser Machine Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805 Abraham Ayegba Alfa, Sanjay Misra, Francisca N. Ogwueleka, Ravin Ahuja, Adewole Adewumi, Robertas Damasevicius and Rytis Maskeliunas Automatic 3D Reconstruction Detection System for Knee Osteoarthritis Based on K-Means Algorithm . . . . . . . . . . . . . . . . . . . . . 819 Kadry Ali Ezzat, Lamia Nabil Mahdy, Aboul Ella Hassanien, Ashraf Darwish, Snasel Vaclav and Deepak Gupta Optimized Twin Support Vector Clustering in Transmission Electron Microscope of Cobalt Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 829 Atrab A. Abd El-Aziz, Heba Al Shater, A. Dakhlaoui, Aboul Ella Hassanien and Deepak Gupta Transfer Learning with a Fine-Tuned CNN Model for Classifying Augmented Natural Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 843 Dalia Ezzat, Aboul Ella Hassanien, Mohamed Hamed N. Taha, Siddhartha Bhattacharyya and Snasel Vaclav Classification of Human Sperm Head in Microscopic Images Using Twin Support Vector Machine and Neural Network . . . . . . . . . . . . . . . 857 Kamel K. Mohammed, Heba M. Afify, Fayez Fouda, Aboul Ella Hassanien, Siddhartha Bhattacharyya and Snasel Vaclav Challenges of Big Data Visualization in Internet-of-Things Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 873 Doaa Mohey Eldin, Aboul Ella Hassanien and Ehab E. Hassanien

xxiv

Contents

Multiple Cyclic Swarming Optimization for Uni- and Multi-modal Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 887 I. Fares, Rizk M. Rizk-Allah, Aboul Ella Hassanien and Snasel Vaclav Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 899

About the Editors

Ashish Khanna received his Ph.D. from NIT, Kurukshetra in March 2017, his M.Tech. in 2009 & his B.Tech. from GGSIPU, Delhi in 2004. He completed his postdoc at Inatel, Brazil. He has published 72 research papers as well as book chapters in reputed journals and conferences. He has also authored/edited 14 books. Deepak Gupta received his Ph.D. in CSE from Dr. APJ Abdul Kalam Technical University, M.E. from Delhi University, & B.Tech. in 2017, 2010 and 2005 respectively. He completed his postdoc at Inatel, Brazil. He is currently working at Maharaja Agrasen Institute of Technology, GGSIPU, India. He published 82 papers in international journals and conferences. He has authored/edited 36 books with international publishers. Siddhartha Bhattacharyya completed his bachelor’s degrees in Physics and Optics & Optoelectronics and master’s in Optics & Optoelectronics at the University of Calcutta in 1995, 1998 and 2000, respectively, and his Ph.D. in CSE at Jadavpur University in 2008. He is currently the Principal of the RCC Institute of Information Technology, Kolkata, India. He is a co-author/editor of 24 books and has published more than 200 papers in international journals and conferences. Vaclav Snasel is Dean of the Faculty of Electrical Engineering and Computer Science at the Mining University, Technical University of Ostrava. He has almost 30 years’ experience in academia and research including industrial cooperations. He works in a multi-disciplinary environment on various real world problems. Jan Platos received his Bachelor of Computer Science in 2005; Master of Computer Science in 2006 and Ph.D. in Computer Science (2010). He is currently the Head of the Department of Computer Science at the Faculty of Electrical Engineering and Computer Science, VSB-Technical University of Ostrava. He is author of more than 178 papers in international journals and conferences. He has also organized 11 international conferences.

xxv

xxvi

About the Editors

Aboul Ella Hassanien (Abo) received his B.Sc. with honours in 1986 and M.Sc. degree in 1993, from the Pure Mathematics and Computer Science Department at the Faculty of Science, Ain Shams University, Cairo, Egypt. He received his doctoral degree from Tokyo Institute of Technology, Japan in 1998. He is a Full Professor at the IT Department, Faculty of Computer and Information, Cairo University. He has authored over 380 research publications in leading peer-reviewed journals, book chapters and conference proceedings.

Improving the Accuracy of Collaborative Filtering-Based Recommendations by Considering the Temporal Variance of Top-N Neighbors Pradeep Kumar Singh, Showmik Setta, Pijush Kanti Dutta Pramanik and Prasenjit Choudhury Abstract The accuracy of the recommendation process based on neighborhoodbased collaborative filtering tends to diverge because the interests/preferences of the neighbors are likely to change along with time. The traditional recommendation methods do not consider the shifted likings of the neighbors; hence, the calculated set of neighbors does not always reflect the optimal neighborhood at any given point of time. In this paper, we propose a novel approach to calculate the similarity between users and find the similar neighbors of the target user in different time period to improve the accuracy in personalized recommendation. The performance of the proposed algorithm is tested on the MovieLens dataset using different performance metrics viz. MAE, RMSE, precision, recall, F-score, and accuracy. Keywords Recommender systems · Collaborative filtering · Similarity metrics · Prediction approach · Rating · Top-n neighbor · Time period · Cluster · MAE · RMSE · Precision · Recall · F-score · Accuracy

1 Introduction Collaborative filtering (CF) is one of the most popular filtering approaches used in e-commerce applications for recommending online items to the users [1]. To predict the recommendable items, CF-based recommendation systems consider the alikeness P. K. Singh (B) · P. K. D. Pramanik · P. Choudhury Department of Computer Science and Engineering, National Institute of Technology Durgapur, Durgapur, India e-mail: [email protected] P. K. D. Pramanik e-mail: [email protected] P. Choudhury e-mail: [email protected] S. Setta Department of Computer Application, Techno India Hooghly, Chinsurah, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 A. Khanna et al. (eds.), International Conference on Innovative Computing and Communications, Advances in Intelligent Systems and Computing 1087, https://doi.org/10.1007/978-981-15-1286-5_1

1

2

P. K. Singh et al.

of the ratings (rating similarity) given by the similar users (neighbors) for an item. Similarity measures play an important role in the accuracy of CF [1]. An inaccurate top-n similar neighbor of the target user creates low prediction accuracy in CF-based recommendation systems [2]. However, traditional similarity measures have certain limitations to find the similar neighbors of the target user in different time periods.

1.1 Problem Definition The similar rating pattern of two users suggests that they might have similar likings. But human preferences change over the time. And, as a result, the list of neighbors of a particular user also changes. For example, Table 1 shows a list of four users and six movies with the rating information. All users provide the ratings of the first three movies in 1996 and the remaining three movies in 1997. The above table represents the changing interest of User 1. The most similar user of User 1 is User 2 in 1996 but his rating pattern is more similar to User 3 in 1997. Traditional methods for finding similarity measures use the complete table. The accuracy of the recommendation based on the old set of similar users tends to decrease along with time. Hence, there is a need of a novel neighborhood calculation approach that considers the changing preferences of the neighbors to enhance the accuracy in personalized recommendations.

1.2 Proposed Solution Approach To include the temporal information in calculating the neighborhood, we have considered user ratings per year basis. The optimized k-means clustering is applied on years to find the cluster of similar users in optimized cluster of years. The philosophy of considering optimized clustering on years is that two users are considered to be more similar if their yearly rating behavior on co-rated items and yearly number of ratings are similar.

Table 1 Rating information of users in different years Year 1996 1997 User Movie A B C D User 1 User 2 User 3 User 4

3.5 3 0.5 0.5

4.5 4 1.5 1

1 1.5 0.5 1.5

2 1.5 1.5 2

E

F

4 2 4 5

3 1 3.5 3

Improving the Accuracy of Collaborative Filtering …

3

The advantage of this approach is that we can find the top-n neighbors of the target users in different time intervals. This will improve the accuracy in personalized recommendation compared to using traditional similarity metrics.

1.3 Advantage of the Proposed Approach The work of this paper offers twofold advantages: – Mitigates the limitation of the traditional similarity measures to find the similar users in different time period. – Leads to more personalized and accurate recommendation.

1.4 Contribution of This Paper The aim of the proposed approach is to extract the similar users of the target user in different time intervals for more personalized recommendations and to improve the accuracy of the traditional neighborhood-based CF algorithms. The main contributions of this paper are as follows: – Calculate the total number of ratings provided by a user in different time intervals. – On the dataset that contains users’ total number of rating information in different years, apply the optimized K-means clustering algorithms (Elbow method) to find the optimal number of clusters. – Comparison of the traditional CF algorithm and the proposed approach on the basis of performance metrics, i.e., MAE, RMSE, precision, recall, and accuracy.

1.5 Organization of the Paper Section 2 discusses the associated background very briefly. Not many works have addressed the temporal information for calculating the neighborhood. Few of them are mentioned in this section. Section 3 presents the solution details which also includes two algorithms: (a) finding optimal number of clusters of similar users and (b) finding the top-n neighbors of a target user. Section 4 does a comparative analysis of the proposed approach on the MovieLens dataset using the performance metrics such as MAE, RMSE, precision, recall, F-score, and accuracy. Finally, Sect. 5 concludes the paper.

4

P. K. Singh et al.

2 Background and Related Work The RS gives probabilistic suggestion for products or items to users on their explicit and implicit preferences, using the help of previously collected data of users and items. Content-based and CF are the two main filtering approaches used in RS [3]. In content-based recommendation [4], RS recommends items that are similar to those items which are liked by a user in the past, whereas CF identifies the top-n similar users that have similar taste to the target user. Breese et al. [5] have introduced the two classifications of CF, i.e., model-based CF and memory-based CF. In a model-based CF, firstly training process is completed using the previous information of users and items and then uses for prediction. Campos et al. [6] have proposed a Bayesian network and combine the features of content-based filtering and collaborative filtering. They provide their effective experimental results on the MovieLens dataset. Similarity between top-n users/items and prediction computation using these top-n similar users/items are the two basic steps in memory-based CF. Based on the user similarity and item similarity, the memory-based CF can be categorized into user-based CF and item-based CF, respectively. Scalability and sparsity are the two major issues in CF. Li et al. [7] have minimized the scalability problem with optimized MapReduce for item-based collaborative filtering recommendation algorithm. To mitigate the sparsity issue, Kanti et al. [8] have introduced a new CF approach that combines both similarities user and item for rating prediction. Lee et al. [9] have elaborated a time-based recommender system for more accurate recommendation. Their proposed system has utilized the temporal information, i.e., user purchase time and item launch time with the rating information to find the more personalized neighbors in different time intervals. Koohi el al. [10] have proposed a method to find similar neighbors of target user to improve the performance of user-based CF. Najafabadi et al. [11] have used the concept of k-means clustering and association rules mining to improve the accuracy of collaborative filtering recommendations. But, k-means clustering has limitation to determine the optimal number of cluster that will provide high accuracy. Hence, our work mainly focuses to minimize the limitation of k-means clustering with more personalized recommendation.

3 Proposed Recommendation Approach Rating provides the users’ feedback or their inherent interest on the particular item. Users’ interest is dynamic in nature so that their similar neighbor set also changes in different time periods. The philosophy of the proposed approach is that the two users are said to be similar if their rating patterns are similar in different time periods. Hence, the proposed work includes a matrix Yuc (users’ yearly contribution) that shows the total number of rating provided by a user in a particular year.

Improving the Accuracy of Collaborative Filtering … Table 2 A matrix of users’ yearly contribution User Year Y1 ... ... Yj U ser 1 ... U ser i ... U ser m

Yuc (1, 1) ... Yuc (i, 1) ... Yuc (n, 1)

... ... ... ... ... ... ... ... ... ...

Yuc (1, j) ... Yuc (i, j) ... Yuc (m, j)

5

... ...

Yk

... ... ... ... ... ... ... ... ... ...

Yuc (1, k) ... Yuc (i, k) ... Yuc (n, k)

Table 2 shows a matrix of size m × k, where m and k represent the number of users and the number of years, respectively. If a user i has rated n items in jth year then the value of Yuc (i, j) will be n. The procedure of finding optimal number of cluster of similar users is shown in Algorithm 1.

Algorithm 1 : Finding optimal number of clusters of similar users 1: Input: Yuc dataset. 2: Output: Optimal number of cluster of similar users. 3: Procedure: 4: For c = 1 to k, compute the k-means clustering algorithm on Yuc , where k represents the total number of years   5: For each c, calculate the SSE (sum of squared error) using kc=1 u∈Cc dist (u, Cc )2 . where C represents a set of clusters C=(C1 , C2 ,..., Cc ...Ck ) and dist is a function that calculates the distance between user u and cluster centroid. 6: Plot the curve of SSE for each cluster c=1 to k. 7: The location of a bend (knee) in the plot where c and SSE value will be low, is considered as the optimal number of cluser Co .

Furthermore, in the proposed approach the following equation is computed to  

recommend the top-n items to a user, rˆui =

k C(u,c) kc=1 (r¯u c=1 |C(u,c)|

+

v∈Ni (u)



sim(u,v)(rvi −r¯v ) ). |sim(u,v)|

v∈Ni (u)

Here, rˆui denotes the predicted rating of target user u on the item i and C(u, c) represents a binary matrix that shows the belonging nature of user u in cluster c. If user u belongs to the cluster c, the value of C(u, c) will be 1 otherwise 0. r¯u and r¯v show the average rating of user u and v, respectively. rvi represents the rating of user v on item i, whereas sim(u, v) identifies the similarity between user u and v. Algorithm 2 represents the complete steps in the proposed recommendation approach.

6

P. K. Singh et al.

Algorithm 2 : Recommendation of top-n list to the target user 1: Input: User-Item rating dataset, A set of users (U), items (I) and Time (Y) when a user gives his rating. 2: Output: A list of top-n items to the target user u based on the predicted rating using proposed algorithm. 3: Procedure: 4: For ∀i ∈ U, ∀j ∈ Y, calculate Yuc (i, j) 5: Apply the optimized k-means clustering algorithm (Elbow method) on the Yuc matrix and compute the optimal number of clusters C=(C1 , C2 ,..., Cc ...Ck ). 6: For ∀i ∈ U, ∀j ∈ I, if Ri j  == 0 then, 7: rui ˆ =

k C(u,c) kc=1 (r¯u c=1 |C(u,c)|

+

v∈Ni (u)



sim(u,v)(rvi −r¯v ) ) |sim(u,v)|

v∈Ni (u)

// Ri j means rating of user u on item i.

8: Generate a list of top-n items based of the predicted rating.

Table 3 Used similarity metric and prediction approach Pearson correlation

sim(u, v) = √ 2 



 Mean centering

i∈I (ri,u −r¯u )(ri,v −r¯v )  2 2 2 i∈I (ri,v −r¯v )

i∈I (ri,u −r¯u )

rui ˆ = r¯u +

√

v∈Ni (u) sim(u,v)(rvi −r¯v )



v∈Ni (u) |sim(u,v)|

4 Comparative Analysis The MovieLens dataset ml-100k has been collected to compare the experimental analysis [12]. This collected dataset has 100,000 rating information of 943 users and 1682 movies. The ratings belong in the range of 1 to 5 with 1 increment, 1 denotes the lowest rating whereas 5 represents the highest rating. The collected dataset that has 93.695% sparsity. Based on different training and testing sets, the collected dataset is divided into Dataset 1 and Dataset 2. Dataset 1 contains 45% trained dataset and 55% test dataset, whereas Dataset 2 has 35% trained dataset and 65% test dataset. We use Pearson Correlation as a similarity measure and Mean Centering prediction approach for rating prediction as shown in Table 3. In Table 3, ri,u and ri,v denote the rating of user u and v on item i. Six different metrics i.e., MAE, RMSE, precision, recall, F-Score, and accuracy have been used for evaluation of the proposed approach [1]. The equations of computing MAE and RMSE values are as follows: N | pi − qˆi | (1) MAE = i=1 N  N 2 i=1 ( pi − qˆi ) RMSE = (2) N

Improving the Accuracy of Collaborative Filtering …

7

Table 4 Classification of the possible results of a recommendation of an item to a user Type of ratings Prediction Recommended Not recommended (Predicted high rating) (Predicted low rating) Actual high rating Actual low rating

True-Positive (t p ) False-Positive ( f p )

False-Negative ( f n ) True-Negative (tn )

Here, pi and qˆi show the predicted and actual rating of item i, respectively. N represents the total number of predicted item. We consider the ratings above 3 as a high rating (recommended items), and less than 3 as a low rating (not recommended items). The classification of the possible results is shown in Table 4. Hence, using Table 4, the equations of precision, recall, F-Score, and accuracy become: Precision =

(3)

#t p #t p + # f n

(4)

Precision ∗ Recall Precision + Recall

(5)

#t p + #tn #t p + #tn + # f p + # f n

(6)

Recall = F − Score = 2 ∗ Accuracy =

#t p #t p + # f p

Here, # denotes the ‘number of’. Figures 1 and 2 show that in case of different similar users the proposed CF outperforms the traditional CF algorithm.

5 Conclusion Collaborative filtering has been the most popular technique used in recommendation systems to suggest online items to the users. The recommendation to a user is done based on the preferences of the similar users (neighbors) who rated the same items likewise. Hence, for personalized and accurate recommendation, it is crucial to find the set of neighbors correctly. With the explosive growth of the number of users, the traditional CF faces difficulty in findings of top-n neighbors of the target user. Especially, traditional similarity measures have issues in computing the top-n similar users due to the changing interest of the target user over time. This affects the accuracy of the recommendation. This paper addresses this problem by calcu-

8

P. K. Singh et al. Comparison based on MAE values

Comparison based on RMSE values

Comparison based on Precision

Fig. 1 Comparison between traditional CF and proposed approach based on MAE, RMSE, and precision values

Improving the Accuracy of Collaborative Filtering …

9

Comparison based on Recall values

Comparison based on F-Score

Comparison based on Accuracy

Fig. 2 Comparison between traditional CF and proposed approach based on recall, F-Score, and accuracy

10

P. K. Singh et al.

lating the top-n neighbors per year basis. This approach guarantees that the list of top-n neighbors always remains updated as per the changed preferences of the target user’s neighbors. As a result, the proposed CF provides significantly improvement in prediction accuracy than the existing traditional CF algorithm.

References 1. S.K. Singh, P.K.D Pramanik, P. Choudhury, A comparative study of different similarity metrics in highly sparse rating sataset, in Proceedings of ICDMAI of Data Management, Analytics and Innovation, vol. 2 (Springer, Berlin, 2018), pp. 45–60 2. A.M. Jorge, J. Vinagre, M. Domingues, J. Gama, C. Soares, P. Matuszyk, M. Spiliopoulou, Scalable Online Top-N Recommender Systems (Springer International Publishing, Berlin, 2017) 3. M. Balabanovi´c, Y. Shoham, Fab: content-based, collaborative recommendation. Commun. ACM 40(3), 66–72 (1997) 4. B. Mobasher, Data mining for web personalization, in The Adaptive Web (Springer, Heidelberg, 2007), pp. 90–135 5. J. Breese, D. Hecherman, C. Kadie, Empirical analysis of predictive algorithms for collaborative filtering, in Proceedings of the 14th Conference on Uncertainty in Artificial Intelligence (1998), pp. 43–52 6. L.M. Campos, J.M. de Fernández-Luna, J.F. Huete, M.A. Rueda-Morales, Combining contentbased and collaborative recommendations: A hybrid approach based on Bayesian networks. Int. J. Approximate Reasoning 51, 785–799 (2010) 7. C. Li, K. He, CBMR: An optimized map reduce for item-based collaborative filtering recommendation algorithm with empirical analysis. Concurrency Comput. Pract. Experience 29(10), e4092 (2017) 8. S. Kanti, T. Mahara, Merging user and item based collaborative filtering to alleviate data sparsity. Int. J. Syst. Assur. Eng. Manag. 1–7 (2016) 9. T.Q. Lee, Y. Park, Y.T. Park, A time-based approach to effective recommender systems using implicit feedback. Expert Syst. Appl. 34(4), 3055–3062 (2008) 10. H. Koohi, K. Kiani, A new method to find neighbor users that improves the performance of collaborative filtering. Expert Syst. Appl. 83, 30–39 (2017) 11. M.K. Najafabadi, M.N. Mahrin, S. Chuprat, H.M. Sarkan, Improving the accuracy of collaborative filtering recommendations using clustering and association rules mining on implicit data. Comput. Hum. Behav. 67, 113–128 (2017) 12. MovieLens | GroupLens: https://grouplens.org/datasets/movielens/. Last Accessed 18 Aug 2018

Exploring the Effect of Tasks Difficulty on Usability Scores of Academic Websites Computed Using SUS Kalpna Sagar and Anju Saha

Abstract The prime objective of this study is to empirically determine the effect of tasks difficulty on usability scores computed using the System Usability Scale (SUS). Usability dataset is created by involving twelve end-users that evaluate the usability of 15 academic websites in a laboratory. Each end-user performs three subsets of six tasks whose difficulty is varying from easy to impossible under six different categories. Results are obtained after applying two statistical techniques, one is ANOVA and other is the correlation with regression. Results show that the SUS scores vary from higher to lower values when end-users conduct usability assessment with a list of easy, moderate, and impossible tasks on academic websites. The results also indicate the effect of tasks difficulty on the correlation between the SUS scores and task success rate. Though, the strength of the correlation is strong with each subset of tasks but it varies that depends on the nature of the tasks. Keywords Usability evaluation · System usability scale · Quantitative assessment · Academic websites · Usability metric · Tasks difficulty

1 Introduction In human–computer interaction, during the last decade, researchers have shown strong interest in the quantitative assessment of usability which is conducted by using various surveys. SUS containing ten questions [1] is a widely used survey that is applied to a variety of interactive systems [2–4], and under various task scenarios [5]. SUS is considered as a powerful tool for computation and comparison of usability score for any interactive system. In the usability engineering research domain, K. Sagar (B) · A. Saha University School of Information and Communication Technology, GGSIPU, Sector 16-C, Delhi 110078, India e-mail: [email protected] A. Saha e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 A. Khanna et al. (eds.), International Conference on Innovative Computing and Communications, Advances in Intelligent Systems and Computing 1087, https://doi.org/10.1007/978-981-15-1286-5_2

11

12

K. Sagar and A. Saha

the effect of tasks difficulty on SUS scores is mentioned as an open issue that motivates us to perform the experiment for this study [6]. So, the aim of this paper is to determine the effect of tasks difficulty, i.e. ranging from easy to impossible types, on usability scores computed using SUS. However, SUS does not provide information about how does one interactive system is more usable than other. Due to this reason, conventional ISO metrics [7] are needed to experiment further to have a detailed picture of system usability [5]. The recent study had also been found that shows the strong positive correlation between subjective usability measures and ISO metric, i.e. the task success rates, for both the laboratory and field studies at the individual and system level [5]. It also motivates us to explore the effect of task difficulty on the correlation between the SUS scores and success rate of tasks ranging from easy to impossible types. The remaining sections of this research paper are organized as follows: Section 2 describes the research methodology. Section 3 explains the experimental results with the analysis. Section 4 contains conclusions with future directions and last section contains appendix.

2 Research Methodology 2.1 End-Users Details Twelve end-users are involved in the usability evaluation process of 15 academic websites. Each end-user of varying age groups frequently assesses the academic websites and is having diverse educational and professional backgrounds.

2.2 Three Subsets of Tasks Considered for Usability Evaluation 15 academic websites1 are evaluated with three subsets of tasks in which difficulty level varies from easy to impossible under six different categories as presented below in Table 1. Also, frequently used tasks by all the stakeholders are only considered and the purpose of defining these tasks is to identify whether the website satisfies the usability goals required by all stakeholders or not.

1 Academic

websites listed in NIRF are considered [8–10].

Exploring the Effect of Tasks Difficulty on Usability Scores …

13

Table 1 List of categories considered for usability evaluation with three subsets of tasks Categories

List of tasks Impossible tasks

Medium-difficulty-level tasks

Easy-level tasks

1. Content

Find online user feedback for the professor

Find the concerned person responsible for uploading upcoming events if required then contact

Find vision and mission statement

2. University program

Find out Ph.D. supervisors details like available slots, etc

Determine how to apply for the program

Find out about various programs conducted in university

3. Navigation

Find the sitemap

Show the users where they are

Navigate to homepage from any navigational level

4. Search

Find screen reader on the website for visually impaired students

Find the image and video-based search results

Find simple and advanced search facility

5. Mobile

Search for last semester result on the mobile device

Determine the website has its mobile app

Determine the university website’s text readable on mobile devices

6. Social media

Find any digital storytelling media to help students

Find Blog of university

Find the Facebook page of the university

2.3 Procedure Each end-user performs all the tasks mentioned in Table 1 on the assigned academic website. After executing these tasks in the laboratory, each end-user is allowed to fill 10-item SUS survey2 on the basis of their task experiences during the experiments. The Google form is generated for filling the responses in the SUS survey. The procedure for computation of the SUS scores is adopted from [1] whose value varies from 0 to 100. Any interactive system having the higher SUS score indicates that it has higher usability rating. Further, an ISO metric of system effectiveness as a task success rate on each academic website is also calculated [5, 11].

2 SUS

Survey is mentioned in Appendix Section as Fig. 4.

14

K. Sagar and A. Saha

3 Experimental Results This section describes the experimental results which are obtained after performing two statistical techniques, i.e. one way ANOVA for comparative analysis, and correlation with regression for investigating the relationship between SUS scores and task success rate.

3.1 Comparative Analysis The comparative analysis is performed showing how the SUS scores vary when usability assessment is done with a list of easy, moderate, and impossible-level tasks. Once the usability dataset is collected by the end-users for 15 academic websites, one way ANOVA, the statistical technique in Minitab 17 is applied. Table 2 contains all the experimental and computed values of SUS mean for 15 academic websites which are evaluated with three subsets of tasks. Results show an overall difference in usability ratings of 15 academic websites between three subsets of tasks as p < 0.001. Further, from the Table 2, it is observed that the end-users executing easy-level tasks have rated the 13 websites with higher SUS scores as compared to other two groups. On the contrary, the end-users executing impossible type tasks have rated the 15 academic websites with lowest SUS scores. Research practitioners must carefully consider the tasks for usability evaluation as its difficulty level affect the SUS scores.

3.2 Correlation and Regression Another statistical technique, i.e. correlation and regression are also applied to the collected usability dataset. The correlation is one of statistical measure that determines association of two variables, whereas the regression explains how an independent variable is numerically related to the dependent variable. The standard Pearson correlation was executed in Minitab 17 to investigate the relationship between computed SUS scores and task success rate of three subcategories of tasks ranging from easy, moderate and impossible tasks. Table 3 contains all the experimental and computed values. Results investigated the positive correlation between SUS scores and task success rate for academic websites for three subcategories of tasks. As seen in Fig. 1(Easy tasks), Fig. 2 (Moderate tasks), and Fig. 3 (Impossible tasks), higher SUS scores are associated with higher success rate and vice versa. In other words, easy-level tasks are having higher success rate and higher SUS scores. For easy-level tasks, r(15) = 0.951, p < 0.001, r 2 = 0.905, for moderate-level tasks, r(15) = 0.914, p < 0.001, r 2 = 0.836, for impossible-level tasks, r(15) = 0.971, p < 0.001, r 2 = 0.943.

Exploring the Effect of Tasks Difficulty on Usability Scores …

15

Table 2 SUS scores mean, standard deviation, F-value, and P-value for 15 academic websites broken down by tasks level difficulty Universities

Easy tasks

Moderate tasks

Impossible tasks

F-value

P-value

1. IISC

M = 59.79, SD = 0.72

M = 37.50, SD = 2.61

M = 23.75, SD = 5.86

283.58

0.000

2.ictmumbai

M = 50.00, SD = 1.06

M = 29.58, SD = 3.34

M = 12.50, SD = 2.61

663.38

0.000

3.jnu

M = 53.75, SD = 1.30

M = 25.00, SD = 5.22

M = 54.79, SD = 3.27

259.06

0.000

4.uohyd

M = 57.50, SD = 1.85

M = 27.50, SD = 2.61

M = 25.00, SD = 5.22

314.00

0.000

5.tezu

M = 51.04, SD = 4.19

M = 40.42, SD = 2.79

M = 29.58, SD = 3.34

113.52

0.000

6.du

M = 58.13, SD = 2.17

M = 29.58, SD = 3.34

M = 31.25, SD = 3.92

296.07

0.000

7.bhu

M = 42.50, SD = 5.44

M = 29.38, SD = 3.39

M = 27.50, SD = 3.69

43.97

0.000

8.iist

M = 36.25, SD = 2.5

M = 28.96, SD = 3.76

M = 26.04, SD = 4.19

26.21

0.000

9.bits

M = 53.96, SD = 4.19

M = 37.50, SD = 2.61

M = 27.50, SD = 3.69

169.00

0.000

10.amu

M = 54.79, SD = 3.28

M = 41.67, SD = 1.62

M = 30.63, SD = 6.92

86.00

0.000

11.visvabharti

M = 41.25, SD = 6.08

M = 25.83, SD = 3.74

M = 32.08, SD = 3.34

34.84

0.000

12.puchd

M = 33.75, SD = 2.5

M = 35.41, SD = 1.44

M = 27.08, SD = 3.51

33.91

0.000

13.pondi

M = 53.54, SD = 4.19

M = 32.71, SD = 2.91

M = 30.83, SD = 2.89

166.38

0.000

14.b-u-ac

M = 53.96, SD = 1.29

M = 33.12, SD = 2.41

M = 28.33, SD = 2.68

455.22

0.000

15.nehu

M = 23.96, SD = 2.91

M = 24.17, SD = 4.44

M = 13.13, SD = 2.17

43.70

0.000

These results indicate a strong relationship between the SUS scores and task success rate, although the strength of this relationship is different. This strength of correlation depends on the nature of the tasks that are considered to establish the correlation. Our findings are echoed with [5, 12]. This research work encourages novice researchers to carefully consider the tasks and their difficulty levels for usability assessment as a determinant in the estimation of usability through SUS scores. One significant question can be asked that is why it is important for researchers to understand the effect of tasks difficulty on usability scores computed using SUS. It is simply because intuitive nature of researchers can lead them to take inappropriate

16

K. Sagar and A. Saha

Table 3 Mean SUS Scores computed with the execution of easy, moderate, impossible level tasks, and task success rate for 15 academic websites Universities

SUS mean of easy tasks

Success rate of easy tasks

SUS mean of moderate tasks

Success rate of moderate tasks

SUS mean of impossible tasks

Success rate of impossible tasks

1. IISC

59.79

98.61

37.5

2.ictmumbai

50

75

29.58

65.28

23.75

16.67

33.33

12.50

1.39

3.jnu

53.75

83.33

25

20.83

54.79

84.72

4.uohyd

57.5

97.22

27.50

22.22

25

18.05

5.tezu

51.04

80.55

40.41

50

29.58

33.33

6.du

58.12

97.22

29.58

33.33

31.25

37.50

7.bhu

42.5

69.44

29.38

33.33

27.50

20.83

8.iist

36.25

65.27

28.96

31.94

26.04

19.44

9.bits

53.95

80.55

37.50

47.22

27.50

20.83

10.amu

54.79

84.72

41.66

51.39

30.63

34.72

11.visvabharti

41.25

68.05

25.83

19.44

32.08

34.72

12.puchd

33.75

63.88

35.41

45.83

27.08

20.83

13.pondi

53.54

83.33

32.71

36.11

30.83

23.61

14.b-u-ac

53.95

81.94

33.13

36.11

28.33

20.83

15.nehu

23.95

51.38

24.17

19.44

13.13

1.39

For Easy level task, this relation was significant at: 0.905. r(15) 0.951, p