Sentiment Analysis using Support Vector Machines with Diverse Information Sources

摘要: This paper introduces an approach to sentiment analysis which uses support vector machines (SVMs) bring together diverse sources of potentially pertinent information, including several favorability measures for phrases and adjectives and, where available, knowledge the topic text. Models using features introduced are further combined with unigram models have been shown be effective in past (Pang et al., 2002) lemmatized versions models. Experiments on movie review data from Internet Movie Database demonstrate that hybrid SVMs combine unigram-style feature-based those based real-valued obtain superior performance, producing best results yet published this data. Further experiments a feature set enriched information smaller dataset music reviews hand-annotated also reported, suggest incorporating into such may yield improvement.