Extracting Opinion Targets and Opinion Words with CenterNP and syntactic-semantic features

摘要

Extracting opinion targets and opinion words from online reviews are 2 fundamental tasks in opinion mining. This paper proposed a novel approach to collectively extract them with centerNP and other syntactic and semantic features. Researching on the models of opinion targets and words on their syntactic and semantic features through HNC theory, this paper found that the opinion targets are some special noun phases (NP), and we named them centerNPs which are made up with conceptual categories g,r,z,pw,ww and v, and their syntactic position must be in chunk GBK. So this paper extracted the centerNPs as the candidate of the opinion targets, and introduced-other syntactic and semantic features such as syntactic position of opinion targets, the conceptual categories and contextual features of opinion targets and opinion words. The experimental results showed that our approach achieved a good performance.