Experimental Evaluation of a Lexicon- and Corpus-based Ensemble for Multi-way Sentiment Analysis
Minh Duc Cao, Ingrid Zukerman · 2012
We describe a probabilistic approach that combines information obtained from a lexicon with information obtained from a Naïve Bayes (NB) classifier for multi-way sentiment analysis. Our approach also employs grammatical structures to perform adjustments for negations, modifiers and sentence connectives. The performance of this method is compared with that of an NB classifier with feature selection, and MCST – a state-of-the-art system. The results of our evaluation show that the performance of our hybrid approach is at least as good as that of these systems. We also examine the influence of three factors on performance: (1) sentiment-ambiguous sentences, (2) probability of the most probable star rating, and (3) coverage of the lexicon and the NB classifier. Our results indicate that the consideration of these factors supports the identification of regions of improved reliability for sentiment analysis. 1