E-commerce sentiment analysis using graph based approach

An efficient sentiment analysis technique for deep understanding of the underlying meaning and sentiment of a given set of text is highly desirable in many application domains. This paper proposes an effective sentiment analysis model that makes use of keyword extraction and polarity assignment techniques using graph based approach for the advanced analysis of mobile handset reviews collected from different electronic commercial sites. Efficient keyword extraction plays a vital role in sentiment analysis and thus has been used in this paper using degree centrality measure. A domain dedicated polarity assignment technique has been used to assign significant and relevant polarity to the most important keywords that can best describe and represent the underlying meaning and opinion or sentiment of the data. The proposed model outperforms some of the existing models.