Jaccard Index Versus Preferential Attachment: A Comparative Study of Similarity Based Link Prediction Techniques in Complex Networks

Abstract

Link prediction is a critical task in network analysis that aims to forecast potential connections between nodes. Numerous methods have been developed to address this challenge, with similarity-based techniques gaining substantial attention due to their simplicity and effectiveness. This research work presents a comprehensive review of two prominent similarity-based link prediction techniques, namely the Jaccard Index and Preferential Attachment. The Jaccard Index measures the similarity between two nodes based on the overlap of their respective neighbors, providing valuable insights into the shared structural characteristics that influence link formation. In contrast, Preferential Attachment leverages the principle that nodes with higher degrees tend to attract more connections over time, making it a widely used tool for modeling the growth of networks. In this study, we conduct a systematic comparison of the two techniques, examining their underlying principles, applications, strengths, and limitations. We highlight the contexts in which each method excels and explore scenarios where their predictive performances may vary. Moreover, we delve into the theoretical foundations behind these techniques, shedding light on the mathematical formulations that underpin their link prediction capabilities.