Research On Mining Product Features And Opinion Words For Web Reviews

With the development of e-commerce, network reviews will inevitably become animportant reference for people to buy products and businesses to improve their service, buthow to dig out useful information from the mass reviews is a challenging work. Productfeatures mining as one of the key technologies of network reviews mining has become animportant research direction in network reviews mining area. This paper proposees a networkreviews-oriented product features and opinion words mining method, the purpose is to extractthe product features in a large number of network reviews automatically.This paper mines the product features and opinion words based on association rulesalgorithm and the degree of property co-occurrence in the network reviews, and on the basisof product feature set extracts the opinion words by the syntactic parser. The main work is asfollow:Creating the common product features list in the process of creating association rulestransaction file to minimize the impact of the Chinese fragment tool for mining results,extracting the nouns and noun phrases as the product feature set though the association rulesalgorithm, and introducing the PMI into the candidate features pruning, we improve the PMIformula to calculate the PMI value between the candidate features and specifiers, filtering thecandidate features which do not meet the threshold; and then mining the infrequent featuresthough the opinion words to supplement the association rules algorithm, obtaining morecomprehensive and accurate product features.On the basis of the obtained product feature set, generating the parse tree by the syntacticparser, extracting the word pairs which meet the SBV dependencies, obtaing the final opinionwords though three-step pruning.This paper selects a review corpus from large Chinese shopping sites, verifies theproposed method in mining the product features and the opinion words, experiment resultsprove the method is effective.