Enhancing the HL-SOT Approach to Sentiment Analysis via a Localized Feature Selection Framework

In this paper, we propose a Localized Feature Selection (LFS) framework tailored to the HL-SOT approach to sentiment analysis. Within the proposed LFS framework, each node classifier of the HL-SOT approach is able to perform classification on target texts in a locally customized index term space. Extensive empirical analysis against a human-labeled data set demonstrates that with the proposed LFS framework the classification performance of the HL-SOT approach is enhanced with computational efficiency being greatly gained. To find the best feature selection algorithm that caters to the proposed LFS framework, five classic feature selection algorithms are comparatively studied, which indicates that the TS, DF, and MI algorithms achieve generally better performances than the CHI and IG algorithms. Among the five studied algorithms, the TS algorithm is best to be employed by the proposed LFS framework.