Sentence-LDA and Deep Learning: A Combined Approach for Implicit Aspect Extraction and Sentiment Analysis

Abstract

The increasing usage of E-commerce market like Ali Baba, Amazon and JD are increasing day by day, such platform offering customers to analyze their services and provide useful information of the service for future customers. Furthermore, companies utilize consumer information to improve their facilities for surviving and compete in the market. This importance refers to the study of user comments or opinions. A previous study in the field of opinion analysis has expected that a review does not contain different sentiments, although, this is not correct. In this research work, we analyze the reviews which contain multiple sentiments. Aspects based sentiments analysis (ABSA) is suggested to report this problem. ABSA aims to recognize implicit aspects in the text, and then conduct the text sentiment analysis. This method is proposed for solving the above issue, the topic models are used with the combination of deep learning and Bag-of-Words approach.