Prominent Aspect Term Extraction in Aspect Based Sentiment Analysis

Abstract

In recent years unstructured text has flooded on the web and today it is a trend to comments, give feedback, share experiences toward products, articles, social issues, multimedia web documents etc. Most of the online social, as well as, commercial platform have started to provide separate space for user reviews in the form of natural text. This electronic data is becoming very crucial and popular in decision making. Nowadays, sentiment analysis is not a onetime put effort rather it is more important to know about different aspects mentioned in the user comments. Aspect-term extraction is one of the critical subtasks in aspect-based sentiment analysis. Most of the aspect extraction approaches are the domain or context-dependent. In this paper, an approach is proposed to identify prominent aspects using contextual and domain-specific information. The study and experimental results on different categories of aspects are investigated on standard review data-sets.