ICISDM2023 Invited Speakers



Prof. Mohamed Zakaria Kurdi

University of Lynchburg in Virginia, USA

Biography: Dr. Kurdi is an Associate Professor of Computer Science at the University of Lynchburg in Virginia, USA. In addition to his Ph.D. in CS, he has an interdisciplinary background in Software Engineering, Cognitive Science, and Linguistics. Before joining the University of Lynchburg, he worked in several institutions in North America and Europe. His research interests are in text and data mining and their applications to areas like intelligent computer-assisted language education, authorship attribution, bioinformatics, and Social Network Analysis (SNA). He authored a two-volume textbook about Natural Language Processing (NLP) that was published in French and English. His recent work on text mining won two best paper awards and a nomination from three different international conferences.

Speech Title: Analysis of Topical Dynamics within a Social Network: application to the Enron dataset 

Abstract: Social networks are becoming an essential part of people’s lives in modern societies. Most of the previous research on social networks focused on understanding their formal properties. Unlike previous works, this paper tries to uncover the semantic aspects of social networks by tagging them with precise, human interpretable topics. It is shown that such tagging helps shed light on several aspects of social networks. For example, at the ego-centric level, it is shown that, through the proposed concepts of topical specialization and balance, topic annotation can help build activity profiles of the users about the addressed topics within the network. Topics have also been shown to help understand the nature of the interactions between pairs of users. Finally, it is shown that topic tagging of social networks can help better understand the seasonality of the topics and the user’s change of focus over time. Besides helping better understand the social network, it is shown that they help improve some applications involving identifying central users and recommending a friend. 

Dr. Songhui Yue

Charleston Southern University, USA

Biography: Dr. Songhui Yue received a Ph.D. in Computer Science in August of 2019 from the Department of Computer Science at the University of Alabama (UA) in Tuscaloosa, Alabama. While at UA, he worked as a graduate research assistant in the office of Institutional Research and Assessment and a graduate teaching assistant for Software Engineering. His thesis advisor was Dr. Randy Smith.
Dr. Yue’s general research interest lies in the area of software engineering and data mining. His research focuses on topics including software engineering for context-aware software, IoT, smart cities, contextual data analysis, automation of data mining, software security, and computer vision.

Speech Title: A Revisit to Context-aware Computing, History, Current Trends, and Future Directions

Dr. Mohammad Hossain

University of Minnesota Crookston, USA

Biography: Mohammad Hossain is an Assistant Professor of Software Engineering and IT Management at the University of Minnesota Crookston, where he teaches various software engineering courses. He earned his Ph.D. from North Dakota State University in 2016. His Ph.D. dissertation title was “Foundational Algorithms Underlying Horizontal Processing of Vertically Structured Big Data Using pTrees.” His research interest includes Data mining, Machine Learning, Software Engineering, Cybersecurity, Algorithm, etc.

Speech Title: Processing Big Data Using Predicate Tree (pTree) 

Abstract: Dealing with large amounts of data, commonly referred to as "Big Data," is a crucial factor in many industries today. However, many data mining algorithms, whether supervised or unsupervised, have become ineffective due to the use of horizontal processing when it comes to time management. Horizontal processing involves data processing row by row, which has proven to be extremely time-consuming. This approach is especially challenging when dealing with Big Data, which is characterized by high dimensionality (number of features) and high cardinality (number of records). To overcome these challenges, our team proposes a vertical approach using predicate trees (pTree) to structure data into columns of bit slices, ranging from a few to hundreds, and processed vertically, column by column. The vertical approach is a more efficient way of processing data as compared to the traditional horizontal approach. To test and compare our vertical approach to the traditional horizontal approach, we conducted experiments using three basic mathematical operations - addition, subtraction, and multiplication - with 10 different data sizes ranging from half a billion bits to 5 billion bits. Our findings show that our vertical approach outperformed the traditional method for all Boolean operations, resulting in a 24% to 96% speed gain across all data sizes. We are confident that our approach is suitable for complex computations in big data applications and can achieve significant speed gains. The vertical approach using predicate trees is a reliable and efficient way of processing data, and we strongly recommend implementing it in industries that deal with Big Data. Our approach can help companies save time and resources, allowing them to focus on other aspects of their business.