A Hybrid Method of Sentiment Key Sentence Identification Using Lexical Semantics and Syntactic Dependencies

摘要

Many news articles in the Internet portal, blog and forums always have their own emotional orientations. Considering sentiment key sentence plays an important role in supervising social trends and public sentiment state, there has been a significant progress in this area recently, especially the lexicon-based method. However, the lexicon-based method totally dependents on lexical semantics and does not excavate the implied syntactic structure. We propose a new method which integrates lexical semantics and syntactic dependencies. And the method performs dependency parsing on the basis of a novel lexicon-based algorithm. Experimental results on COAE 2014 dataset show that this approach notablely outperforms other baselines of sentiment key sentence identification.