COMMUNITY DETECTION TECHNIQUES AND THEIR APPLICATIONS IN RECOMMENDER SYSTEMS

Abstract

The goal of community detection in network analysis is to identify densely connected groups of nodes with sparse connections between them. This thesis provides a compre hensive exploration of community detection techniques and their applications, with an emphasis on recommender systems. It focuses on the implementation and comparison of three community detection al gorithms: the Louvain Algorithm, K-means clustering Algorithm, and Gaussian Mixture Model. A supply chain dataset is utilized as the basis for experimentation, allowing for the identification of communities within the network structure. Analysis and evaluation of algorithms’ performance offer insights into their strengths and limitations, offering a comprehensive understanding of their effectiveness in detecting communities within the supply chain domain. It also offers a comprehensive review of community detection approaches, highlight ing their applications across various domains. The literature review explores different algorithmic approaches, including modularity-based methods, hierarchical clustering, and graph partitioning algorithms. The strengths, limitations, and potential applications of these techniques are discussed, providing valuable insights for researchers and practition ers interested in community detection. The findings from the implementation and comparison of community detection algo rithms on the supply chain dataset, coupled with the comprehensive review of community detection approaches, contribute to the advancement of knowledge in community de tection. The thesis sheds light on the effectiveness of different algorithms in detecting communities within complex networks, specifically focusing on the supply chain context. The insights gained from this research can aid in understanding the underlying structure and dynamics of networks, enabling more informed decision-making processes. In summary, this thesis provides a comprehensive investigation into community detec tion techniques and their applications. By exploring the implementation and comparison iv of various algorithms on a supply chain dataset and conducting a thorough review of community detection approaches, this research contributes to the existing body of knowl edge in network analysis. The insights and methodologies presented in this thesis can be leveraged by researchers and practitioners in various fields to gain a deeper understanding of community structures within complex networks.