Aspect based sentiment analysis in social media with classifier ensembles

Abstract

The analysis of user generated content on social media and the accurate specification of user opinions towards products and events is quite valuable to many applications. With the proliferation of Web 2.0 and the rapid growth of user-generated content on the web, approaches on aspect level sentiment analysis that yield fine grained information are of great interest. In this work, a classifier ensemble approach for aspect based sentiment analysis is presented. The approach is generic and utilizes latent dirichlet allocation to model a topic and to specify the main aspects that users address. Then, each comment is further analyzed and word dependencies that indicate the interactions between words and aspects are extracted. An ensemble classifier formulated by naive bayes, maximum entropy and support vector machines is designed to recognize the polarity of the user's comment towards each aspect. The evaluation results show sound improvement compared to individual classifiers and indicate that the ensemble system is scalable and accurate in analyzing user generated content and in specifying users' opinions and attitudes.