Weighted Bi-directional GRU Capsule Ensemble Approach for Multi-Domain Sentiment Analysis

With the advent of the Web today, users' opinions can be incorporated into a variety of applications. Automated methods have been developed to derive users' general sense from these textual comments, often known as sentiment analysis, and aim to determine the polarity of a text relative to a subject. One of the challenges is the inability to use one domain of data to analysis sentiment in another domain and the lack of sufficient labelled data in a particular domain. To address these challenges, multi-domain sentiment analysis systems have been developed. This paper propose Bi-GRU Capsule ensemble approaches for multi-domain sentiment classification to address the mentioned issues. Using a weighted score of Term-Frequency and Inverse Document Frequency degree and the initial polarity of the sample test data on each domain, a new aggregated score of final polarity is obtained. The DRANZIERA protocol is used for evaluation of the proposed model. The outcomes demonstrate the effectiveness of the proposed approach and also set a plausible starting point for future work

Keywords:

Multi-domain sentiment analysis , Deep learning , Weighted neural network , Natural language processing , Ensemble method