The Research Of Product Feature Extraction In B2C Website

As the development of B2C electronic commerce, more and more products are selling online, and product information is becoming more detailed. But the problem of information asymmetry between online sellers and buyers still remains and will have negative influence on online transaction and customer satisfaction. Web2.0 has provided convenience to online communication. Using interactive methods of B2C website, sellers can collect customers'opinion and improve product description.In this research, we first explained the information asymmetry problem in electronic commerce, summarized the common interactive methods of B2C website, and proposed that online seller could mine customers'concerned features from the interactive data. Then, through the literature analysis, we introduced the existing algorithm of product feature extraction and proposed the flow of product feature extraction method based on association rule. The flow contains four steps: data preprocessing, construct transaction file, apply Apriori algorithm and optimize frequent items to get final product feature set. We also took experiment of this method by using real customer data on amazon.cn. The experiment result proved that this method is valid, but also there are many problems. For example, the segmentation tool and product category have great influence on the result. At the end of this research, we pointed out the lack of our method and the possible direction of future study on product feature extraction.