Abstract

To improve the performance in e-commerce markets, big giants like Amazon, Myntra and Flipkart are providing consumers with a platform to review their services and also give them an opportunity to provide a useful insight of the service to the future buyers. On the other hand, companies use such reviews to make a significant upgradation in their products (or services) to survive in the competition from others in the market. This shows the importance of studying user views or opinions on a particular product (or service) consumed by users. In Natural Language Processing (NLP), the process of studying such user opinion is termed as opinion mining. It is a task of finding out overall sentiment present in a review. Past research in this area has assumed that a sentence cannot have multiple sentiments associated with it. However, this is not true. For example, “This car looks beautiful, but does not handle very well.” comprises a positive sentiment towards the looks of the car but a negative sentiment towards its handling. To address such issues, aspect-based sentiment analysis (ABSA) was introduced. ABSA aims to detect an aspect (i.e. features) in a given text and then perform sentiment analysis of the text with respect to that aspect. The chapter aims to discuss the concept of ABSA for the problem introduced as a FiQA 2018 challenge subtask 1 (https://sites.google.com/view/fiqa) in WWW 2018 shared task. It highlights all the state-of-the-art models in the domain and discusses some new approaches. We propose neural network models combined with hand-engineered features and attention mechanism, to perform ABSA on financial headlines and microblogs. Our proposed model outperformed the existing state-of-the-art results in sentiment part by 50% and in the aspect part by 20%.