A Text Extension Algorithm Based on Synonymy Discovery
Bo Li · 2014
Nowadays,in the proceeding of text categorization,document which needs to be classified is hardly acquiring features to represent document because of limited terms,which lead low accuracy of text categorization. An improved algorithm of text extension algorithm based on synonymy discovery is proposed,we use the hierarchical architecture in HowNet to get relationship among primitives,obtaining the position of term in pre-classification text and discovering all synonyms in different levels in the path of the term,the corresponding correlation coefficient between term in pre-lassification text and synonyms in different levels is provided through level and density message among primitives. The experimental results on open benchmark datasets of 20Newsgroups and Reuters21578 Top10 show that our approach acquires a high accuracy and better F1performance compared to conventional algorithm.