Real-Time Aspect-Based Sentiment Analysis on Consumer Reviews

Abstract

The rise of e-commerce websites, as new shopping channels, led to an upsurge of review sites for a wide range of services and products. This provides an opportunity to use aspect-based sentiment analysis and mine opinions expressed from text which can help consumers decide what to purchase and businesses to better monitor their reputation and understand the needs of the market. Aspect-based sentiment analysis (ABSA) is a technique aimed to foster research beyond sentence or text-level sentiment classification. The goal is to identify opinions expressed about specific entities (e.g., laptops) and their aspects (e.g., price, performance, build quality, etc.). There exist very few techniques which can generate such results based on customer ratings, however usually for a limited set of pre-defined aspects and not from free-text reviews. The other challenge in this process is cold start problem because of the lack of enough review data for a product. In this paper, a methodology is proposed to automatically compute sentiments of dynamic aspects from user-generated reviews from the web scraping from multiple sources to overcome the cold start problem. Therefore, this methodology is devising a better solution for understanding sentiments in e-commerce than existing methods.