Study Of E-commerce Website Personalized Recommendation Based On Web Text Mining

The purpose of Web text mining is to extract useful information and knowledge from a Web page, and personalized recommendation can be seen as a form of "seek someone by information". Effective information is the basis of generating accurate personalized recommendation, so web text mining has been become the hot research of a personalized recommendation. At the same time, the era of big data is coming, data information has an explosive growth. It is because of this kind of explosive growth, people find it is hard to get themselves interested in this website. So personalized recommendation technology become more important a lot in the field of the Internet.The research of e-commerce personalized recommendation for users and the website owners is a win-win pattern. As for users, it can help them get their needs quick and convenient, and to satisfy their inner needs. For the site owners, it can help them fully display their goods, and can fully mining the long tail goods to sale. It has a meaningful prospect from the value of the research.In this paper, the research on e-commerce personalized recommendation is based on Web text mining model, the data source for recommendation is from the site’s web server logs, and paper also gives the e-commerce personalized recommendation model based on web text mining. First of all, the model analyzes the topology of the e-commerce site and page structure, based on log data filtering, user identification, session identification and path added to extract the transactions. Second, for Web text mining, mainly involves the Chinese word segmentation, feature items weight calculation and the extraction of feature items collection and so on, then convert the transaction set to the characteristic representation and cluster analysis. Finally, when users visit the web site, through the analysis of the characteristics of the current user session information, similarity computing and clustering focus on transactions, and create personalized recommendation sets. The paper established a personalized e-commerce recommendation system based on Web text mining, is used to verify the model, validation of measure for the study of the personalized recommendation recognition in the field of precision rate and recall, recommend the coverage, and novelty. At the same time, the experiment compared the traditional web text mining which is used in TF*IDF algorithm with the BM25F which is choose by paper’s model, analyzed the advantages and disadvantages of the algorithms. In this paper, a personalized recommendation of web text mining is built, it is good for tracking the user interest changes in access page, when a user from the kind of commodity to jump to another item, to extract the feature items will also change, so to be able to timely adjust the recommended set. Figure13table4reference51...