Semantic Related Feature Analysis and Dynamic Evolution Based on Topic Temporal Chains under the Social Network

As traditional topic detection under the online social network is lack of mining the topic semantic features and dynamic evolution characteristics, this paper studies the semantic features and dynamic evolution trend based on topic temporal chain under the online social network. According to the temporal relations, text flows are divided into the text set in continuous time slice, then combination of cosine similarity and Jaccard similarity coefficient is used to implement semantic related feature analysis of topic temporal chain. Experiments on the two real social network are made to verify the topic evolution performance of the proposed method by comparison with the existed CW-TE(Coword-based Topic Evolution) method and RE-TE (Relative-Entropy-based Topic Evolution) method, which demonstrate that the proposed online topic evolution algorithm can more effectively detect semantic related feature and dynamic evolution of topics and obtain better detection performance.