A Dynamic Clustering Method of Hot Topics Based on User Interaction and Text Similarity

Abstract

This paper proposes a dynamic clustering method for hot topics based on user interaction and text similarity. It focuses on the analysis of the clustering process from the perspective of movement and combines the two aspects of text similarity and user interaction to comprehensively consider the topic clustering of microblogs, improve the accuracy of clustering. The simulation results demonstrate that the clustering process is dynamic and can be displayed intuitively. Moreover, the model has strong extensibility, which parameters can be added, deleted and changed according to individual needs, and can be personalized for various applications.