A sentiment analysis hybrid approach for microblogging and E-commerce corpus

Exploiting linguistic features is necessary on sentiment analysis in natural language processing. This paper proposes a novel approach on exploiting linguistic features and SVMperf based semantic classification. The innovation is that it uses the dependency relationship to do the linguistic feature extraction. In order to reduce the computational complexity, this paper uses the X2 (chi-square) and Pointwise Mutual Information (PMI) metrics for feature selection. Furthermore, as for the approach on sentiment analysis, this paper uses the SVMperf based algorithm to do the alternative structural formulation of the SVM optimization problem for classification. This paper uses two different corpuses (i.e., microblogging and e-commerce data set) to evaluate the performance. Experiment results show the feasible of the approach. Existing problems and further works are also present in the end.