User reviews based product feature mining of mobile phones in E-commerce

Product review mining aims to quickly extract useful information from massive comments published by users and adopt an intuitive way to help consumers make purchasing decisions. Fine-grained product feature mining is very important, however, the product characteristics semantics (upper and lower characteristics, synonymous features) analysis is inadequate on existing product reviews researches. Firstly, the ontology of mobile phone features was constructed based on mobile phone descriptions. Then crawling programs was employed to get product comments and followed by conducting words segmentation, part of speech tagging, getting rid of the repeats and other pretreatments. Using the Apriori algorithm, the appropriate product features from user's perspective were extracted. Combining with HowNet dictionary, semantic extension was carried to improve the ontology of product features, which will facilitate further accurate sentiment analysis of the product reviews.