Ranking Product Aspects Based on Consumer Reviews

Abstract

The Internet has become an excellent source for gathering consumer?s opinions or reviews. For product numerous consumer reviews of product are available on internet .Consumer reviews or opinions are useful for both firms & users as they contain rich & valuable knowledge about product. The business firm needs different reviews of customers for development of product. The user can make wise purchasing decision by looking at customer reviews. There are reviews on various aspects of the products. The reviews are numerous, diverse and not precise leading to difficulties in information gathering and knowledge acquisition. A product may have hundreds of aspects. Some of the aspects are important than the others. Therefore we are developing the system to mine those aspects and rank them which will help for better product development. This proposed method is named as ?A product aspect ranking framework?. Among reviews of consumer for particular product, it first identifies aspects in the reviews by a shallow dependency parser and then analyzes consumer opinions on these aspects via a sentiment classifier. Then a probabilistic aspect ranking algorithm is used, which effectively exploits the aspect frequency as well as the influence of consumer?s opinions given to each aspect over their overall opinions on the product in a unified probabilistic model.