Feature Extraction Using Semantic Similarity

Recently, opinion mining has become a focus in the field of natural language processing and web mining. Due to the massive amount of users' reviews on the web about some entities or services, opinion mining is appeared to track users' emotions and feelings. Sentiment analysis is a synonym to opinion mining. Feature extraction is an important task in the sentiment analysis process. So, in this paper, a novel model is proposed to extract the most related features to a product from customer reviews using semantic similarity. Wordnet taxonomy and Stanford Part of Speech (POS) tagger are used in the feature extraction process. The extracted features are very important to generate a meaningful feature based product reviews summery which helps the customers to make a decision. The experiments are performed on three different datasets. The proposed model achieves promising results in terms of Precision, Recall and F-measure performance measures.