A Text Mining Approach to Online Reviews for Opinion Mining Using Ensemble

摘要

Online product reviews are considered as a major information resource which is useful for both customers and manufacturers. The online reviews are unstructured-free-texts in natural language form. The task of manually scanning through huge volumes of review is very tedious and time consuming. Therefore it is needed to automatically process the online reviews and provide the necessary information in a suitable form. In this paper, we dedicate our work to the task of classifying the reviews based on the opinion, i.e. positive or negative opinion. This paper mainly addresses using the ensemble approach of Support Vector Machine (SVM) for opinion mining. Ensemble classifier was examined for feature based product review dataset for three different products. We showed that proposed ensemble of Support Vector Machine is superior to individual baseline approach for opinion mining in terms of error rate and Receiver operating characteristics Curve.