Aspect Extraction in Sentiment Analysis Based on Emotional Affect Using Supervised Approach

Abstract

Aspect based sentiment analysis (ABSA) is regards as the most demanding area of research in the field of text mining and natural language processing. In the last decade, people tend to be more interested in consulting online media outlets to make decisions concerning products or events. However, the existing ABSA approaches alone are not sufficient in providing comprehensive information required to make an informed decision. Most of the early studies focused on the binary classification of “positive” and “negative” to predict an overall sentiment of a review, which reduced the effectiveness in decision making towards the entities. Even though earlier studies on SA are centered on aspect extraction, there are limited works that effectively combine aspect extraction with emotional affects such as surprise, sadness and happiness, which consequently leads to incomplete meaning of a content. In this paper, we investigate a problem involving aspects in relation to emotional affects. To propose a novel supervised approach for the aspect extraction with their associated emotion classes in sentiment analysis (SA), we conducted an analysis of some prominent ABSA approaches. This research work will lead to a more accurate results of SA and could assist both novice and prominent researchers in comprehending the relationship between aspects and emotions in SA considering its sematic role in business intelligence.