A Joint Sentiment-Topic Model for Product Review Analysis of Electronic Goods

Online product review plays an important role in raising the voice of customers and their purchase decision. However, a conventional method in the area has largely focused on field research and surveys to acquire information about customer preferences. The already existing frameworks largely focus on supervised learning model techniques which is not feasible in all cases as it requires a labelled corpus which might be difficult to gather at times. This paper proposes a new framework for a review of topic modelling named joint sentiment topic model. The proposed research work extracts key features of online reviews and present a detailed analysis of different approaches and techniques, comparing the reliability of diverse systems among them. Also, this research work attempts to implement topic modelling in order to find how much a customer’s occupation and interests match a particular product and accordingly give weightage to their reviews. This idea is an effort to explore different approaches and techniques of topic modelling and provide valuable executive implications for electronic goods.