An approach for Aspect Based Sentiment Analysis using Deep Learning

In this project, we present a deep learning approach for aspect based sentiment analysis (ABSA). Sentiment analysis is an important task in natural language processing and has a lot of applications in real world. The typical sentiment analysis is a process of classifying opinions expressed in a text as positive, negative or neutral. A more general task would be to predict the sentiments of each aspect mentioned in the text. In recent years with the growth of internet, social networking sites, discussion forums and blogs, e-commerce websites have gained immense importance. To enhance customer shopping experience, these websites generally provide platform for people to express their views about the product and its different aspects. There is a review and an overall score available for each product, but this doesn’t provide the complete information. For example, two products might have same overall rating but with different unsatisfactory aspect opinions. This problem has inspired a new line of research on aspect level opinion mining since early 2000s [2]. Also, a major source of encouragement behind this project comes from the increasing popularity of ABSA task present in SemEval since 2014.