A Semantic Representation of Micro-blog Short Text Based on Topic Model

摘要

The fundamental challenge confronting micro-blog short texts is to represent, compute and mine their semantics since they are colloquial language and restricted within 140 words. This paper presents an approach to represent micro-blogs semantically based on topic model. First, we get the topic vector of micro-blog through word-themes Mutual Reinforcement of micro-blog texts. Second, we take advantages of improved K-means to cluster micro-blog topic vectors and gain a category set of micro-blogs. Third, we make experiments to prove the above-mentioned methods. Finally, we find these methods can calculate semantics of micro-blogs text accurately and effectively.