Opinion Targets Identification Based on Kernel Sentences Extraction and Candidates Selection

摘要

With the developing of the Internet, communication becomes more and more frequent, and the traditional opinion mining technology has been unable to meet the people's needs, especially in the field of opinion targets identification. Therefore, how to do appropriate pre-processing and postprocessing with opinion sentences to improve the quality of opinion sentence identification has become a hot issue in recent years. Researches on kernel information filtering and candidates screening of traditional opinion targets identification methods are insufficient. In this paper, we propose a novel opinion targets identification method which integrates kernel sentences extraction with candidates selection based on rules analysis and SVM screening. Experimental results on COAE2014 dataset show that this approach notably outperforms other baselines of opinion targets identification.