The Research On Personalized Search System Based On User Interest Model

The rapid development of the Internet not only provides people with large amounts of information, but also makes it difficult for users to obtain valid information quickly. The search engine has become a major way of access to network resources. In practical applications, even if different users, who have different needs, enter the same query word will get the same results, or even the same page sort results. Therefore, the personalized search based on user interest model is a development direction of the search engine.In this paper, through the analysis of personalized search engine framework and the basic technical, designs and implements a prototype system of personalized search based on user interest model. First, this system creates a user interest model using user’s browsing history pages. The system needs to preprocess and Chinese word segement these web pages, uses TF-IDF algorithm to statistics keyword frequency, then selects top-N keywords on weight using space vector representation as feature vector words.Using this model, we can filter the search results and calculate the relevance to remove the uninterested information. Thus we can pick out a high degree of pages that match the query to ensure the relevance of web content. Then, by analyzing the behavior of the user clicks on search results to extract the search results for the query keywords, thus increasing the weight of relevant pages, reducing irrelevant page weights. Finally, integrate the related scores of two phases. According to the scores, this system can filter and reorder the search results pages to realize a single-user-oriented personalized search.Finally, through analysis the user’s browser date and click behavior, design and implement a personalized search system with Nutch and Sorl. During the experiment, with unrealized personalized search system for comparison, the experimental results show that:based on user interest model and click behavior, this personalized search system will re-sort the search results. These results are in line with user’s query needs; effectively improve the system precision rate and user’s satisfaction.