The Research For The Method Of Mining Preference Based On User Behavior And Feedback

It is specially meaningful for service providers to make service providers and service users the mutual interaction between further ascension in the market environment competition under the trend of rapid development in the digital economy. Service providers are facing such a problem, that is how to make service to each user to provide specific goods or services. Personalized service is the best method to solve the problem. Personalized service purchases information through the collection and analysis of the user and user’s preference in learning, pushing by providing high quality services, cultivating loyal customers and attracting new users. By a customers personalized recommendation system needs not only master recommended technology, but also need high accuracy of user data. Select the user preference model, in this way, the system can very good understanding of user interests and preferences and all kinds of user information, is the foundation of personalized service. Users in the network shopping of all kinds of behavior and feedback are found in the user’s interest preference, get high quality user preference information, is the one of the key problems of personalized recommendation system. This paper is mainly based on the user shopping web browsing behavior, the user to commodity review information and rating information fully mining user for web service preference, form a complete user preference model, and use the case study proved that the method can guide and achieve a complete set of user preference mining model, which is used in particular field. This paper first introduces the establishment of user web service preference model is recommended for individual system to provide preference based on the great significance. In simple description of web service user preference model definition and the development present situation, the user web service preference model technology, including the user web service preference knowledge acquisition, the user web service preference modeling technology, web service user preference model, on the basis of representation method is proposed based on hierarchical vector space users web service preference mining model, including the user web service preference modeling scheme, based on hierarchy vector space users web service model expression and model update. The model can fully describe the user preference level relationship, model update can adjust the user interest preference accuracy. The greatest degree of the user a precise statement of commodities types and specific commodity type of attribute preference. This model can be personalized recommendation system for users to recommend reasonable preference goods, in order to improve the accuracy recommended. In the specific user web service preference after model, in this paper, the multiple attribute decision making method to mining user web service behavior preference. On the use of multiple attribute decision making based on the user behavior preference to dig the advantages, as well as to the multiple attribute decision making are summarized, describe the attribute decision making theory foundation, this paper introduces the decision attribute decision making method is the most important method, on the basis of simple weighting method, this paper presents a simple weighted method of inverse process-reverse weighting method. In this paper, according to the user network shopping behavior get goods ranking, according to the idea of reverse weighting method for ranking goods user preferences of digging, users receive goods attribute weight value range, also is the user preference information of commodity. Users in the shopping web site after the purchase of goods for goods simple comments and score. In this paper according to the user’s review score to mining user to purchased goods preferences. First in this paper, the user comments do the common goods are briefly analyzed. Then puts forward comments on goods attribute words and ideas word extraction method, this article with the Chinese academy of sciences institute of computing technology, the developed Chinese grammar analysis system to review sentences ICTCLAS word segmentation and part of speech analysis, with the nearest neighbor matching algorithm for attribute words and ideas word. Then we to view word polarity analysis, studies the common polarity analysis method based on HowNet, mainly introduced the polarity analysis method, this paper puts forward the polarity based on HowNet analysis method, and the improvement of the analysis of the superiority of the improved algorithm. We can represent user’s preferences in the form of linear constraint according to the view word polarity and user rating. Combining\the linear constraint and the conclusion in the third chapter,we can get a new linear programming and the weight of the new value range, it’s also a new updating form of the user’s preference model.