A Local Community and Resource Allocation Perspective on Link Prediction: Method and Visualization

Link prediction is a challenging and complex research area in network studies, which plays an important role not only in recommendation systems in fields such as social media and e-commerce, but also in network security monitoring and biological relationship research. Over the past two decades of link prediction development, many branches in the field of computer science have used their own methods to solve link prediction problems, but analysis methods based on local similarity have stood out due to their lower time complexity, higher prediction accuracy, and interpretability. In this work, a link prediction method based on local community and resource allocation theory is proposed and extended within 3-hop paths, and is also extensively experimented. Compared to most methods based on local similarity, this method has a higher prediction accuracy. In addition, we propose an interactive visualization system, where users can select different indices and datasets to get link prediction recommendations and explore the analysis process.