Bidirectional Feature Transfer for Cross-Domain Sentiment Analysis

With the evolution of user-based web content, people naturally and freely share their opinion in numerous domains. However, this would result in a massive cost to label training data for many domains and prevent us from taking advantage of the shared information across domains. As a result, cross-domain sentiment analysis is a challenging NLP task due to feature and polarity divergence. The main aim of this work is to automatically create a bidirectional thesaurus which could be used to transfer feature vectors of the source and target domains. This paper aims at designing an algorithm of feature transfer to select and transfer the informative and representative features between the source and target domains. Furthermore, several experiments were conducted in order to evaluate the proposed model, and the results were compared to similar known baseline methods.