Sentiment classification using summaries: A comparative investigation of lexical and statistical approaches

Online reviewing has been on the rise and is extremely useful and accessible to web users due to the rise in social networking and more reviewing platforms. Being that some reviews tend to contain more text than is necessary to convey the sentiment of the review, review summarization with respect to polarity classification has become necessary. This work gives a comparative investigation of three forms of summarization approaches used for polarity classification. These include using SentiWordNet for a lexical approach, SVM Light for a statistical approach, and the Open Text Summarizer for a traditional summarization based approach.