Ontology Based Aspect Oriented Opinion Summary Methodology

Abstract

Enterprises selling over the web generally request feedbacks from customers on their purchased products/services. Thanks to e-commerce in an increasingly developing structure, the number of customer feedbacks grows rapidly. Due to the great increase in the number of e-commerce enterprises and customers, it becomes difficult for a potential customer to read these feedbacks while making decisions. It becomes almost impossible for the producer to monitor these feedbacks as well. Product feature extraction from customer reviews is an important operation in opinion mining. The extracted features help to assess the opinions written by customers who have purchased specific products and they provide opinions of customers regarding their positive/negative experiences. Because most of the customer reviews are asyntactic plain texts, methods should be developed for extraction of implicit and explicit product features expressed in customer reviews and comments. In this research, we aim to develop a methodology that examines the opinions, thoughts and comments, written in Turkish, about the products and summarizes the features of the products and the emotional processes related to these properties. Our study differs from others in that product characteristics expressed using synonyms or word groups can be identified, and features can be detected with greater accuracy using ontologies that include product specifications. Our study offers successful results in product feature extraction. Experimental work on opinion/idea/comment for some products on e-commerce sites shows that our methodology yields positive results.