Research on Topic Recognition Based on Multilayer Relation Fusion
Poster sessions are particularly prominent at academic conferences. Posters are usually one frame of a powerpoint (or similar) presentation and are represented at full resolution to make them zoomable.
In this poster, we present a review of the current research status of multi-relational fusion and systematically summarize the multiple relationships among different measurement entities and entities in the scientific literature. Further, we propose a multi-relational extraction and relational fusion approach to thematic identification. We divide the relationships for topic recognition into three types—basic, enhancing, and supplement—that can be formed by integrating co-occurrence, citation, and co-authorship relationships. Finally, as an empirical analysis, we use the PathSelClus algorithm to realize topic clustering based on multivariate relation fusion. Empirical analysis confirms that multivariate relational fusion can effectively improve the effectiveness of topic clustering.