Existing topic tracking methods are mostly fornews and forum data, which lack of statistical methods formicroblogging on relevant topics. Combined withcharacteristics of micro-blog information, the paperproposes a microblogging statistical methods based onsemantic similarity. Firstly by building topic semanticmodel and then use the HowNet semantic similaritycalculation of two terms, and measures the relevance of thetopic and microblogging. Finally statistics method isprovided on the degree of correlation. Experiments showthat novel method works on the problem soundly.
展开▼
