AN APPROACH TO ASPECT DETECTION AND GROUPING FOR OPINION MINING

Abstract

Mining opinions from online reviews is a well-known field. This work proposes a new approach to automatically detection and clustering of aspects and their opinions from online reviews. Clusters of aspects can be used to generate meaningful summaries from reviews, for which, words and phrases of aspects, which are domain synonyms, should be clustered into feature groups. The challenge is to automatically find major themes from the reviews. Supervised methods are not practical due to the diversity in the products and businesses and lack of labelled data. We provide an unsupervised method of grouping reviews which are on the same themes like food quality, decor, service quality etc. Our method uses a combination of semantic similarity methodsdistributional similarity, co-occurrence and knowledge base based similarity. It is shown that opinion words and the context provided by them can prove to be good features for measuring the semantic similarity and relationship of their product features.