Study of Semantic Models in Identifying Aspects in Customer Reviews

Mily Lal, Akanksha Goel, Amruta Chore

Abstract: With the outburst of ecommerce sentiment-rich resources such as online review sites and blogs, people actively use this information to understand what others think about a particular subject. This area of study helps to derive the opinion, sentiment or the outlook of a speaker mainly used when conducting market research. This paper does an evaluation of the systems such as LSA and PMI that set up semantic association between aspects and opinions found in customer reviews. PMI-IR is predicted to give better results as observed in a user study.

Keywords: aspect extraction, opinion mining, opinion word, Sentiment analysis