Deep learning for target-dependent sentiment classification in social media

There is currently a revolution in developing deep learning models for improving performance of many machine learning systems. This revolution has been expanded to sentiment analysis (opinion mining) as a promising research area. Throughout this paper, we swaged in various recent works that are performed for developing sentiment analysis systems by exploiting capabilities of deep learning models. We introduce a comprehensive literature review on predicting sentiments expressed towards particular topics (targets) in micro-blogs (such as tweets). This micro-specialization is referred to as target-dependent sentiment classification. To make our work more comprehensive, we evaluated also two more deep learning models that have not been used before in this research direction. Experimental results are shown along with summaries and discussions to emerge significance of developing deep learning based models in improving accuracy of target-dependent sentiment classification. Our findings highlighted some gaps that can be filled in future research and illustrates that there is a room for improvement.