Paper Title
A Target-Dependent Sentiment Analysis Method For Chinese Micro-Blog
Abstract
Recently with a large number of people publish Micro-blog to share their opinion, Micro-blog becomes an
important data resource for opinion mining and sentiment analysis. The traditional methods usually get poor accuracy since
they ignore structured semantic information and the target of sentimental expressions. In this paper we propose a novel
target-dependent sentiment analysis method. This method obtains the structured information from syntax tree using the
convolution tree kernel of Support Vector Machine. The syntax tree is then pruned according to target with the help of domain
ontology and syntactic paths library. Using this method we can eliminate the effect of irrelevant appraisal expressions.
Experimental results on two corpuses with different targets show that the performance of our method is much better than the
traditional methods.
Keywords- Pruning Strategy, Sentiment Analysis, Syntax Tree, Tree Kernel