A Personalized Search Algorithm By Using User's Preferences Mode

With the change of enterprise production method and the pattern of project research and development, with the acceleration of the construction of enterprise information, within the building and the expansion of a variety of information resource library for enterprise and universities, digital resources grow by the exponential. Traditional information retrieval technologies have high search precision and wide area to meet the needs of some people, but because it is the universal and the public, there is no difference to each user's search results, so the information can not be customized for each user. In the flood of information within the enterprise, the employees or users on how to more quickly locate the information they need and to find the most interesting information to their own, personalized search this paper discussed is to solve this problem.This paper analyzes the traditional user's preferences model, for the lack of traditional preferences model, the authors propose a model with the weighted interest tree to express user's preferences, the model combines the probabilistic model and vector space model, to effectively express the user's preferences; the use of the preference factor of the field classification model to express the user's preferences, the introduction of forgetting factor to optimize the adjusted algorithm of the user's preferences models. The search results are obtained by the search engine are often very large, the success of search engine is determined by filtering and sorting of search results. And how to filter these results document, filter out the closest article to the user's interest, is an important issue to be soluted by personalized search engine. Based on the deficiencies of traditional filtering methods, this paper proposed the concept of area of sorting and partial sequencing and made a personalized, integrated sorting algorithms.Finally, the design and implementation enterprise resource management system is given, and in this resource management platform we design the Personalized Search modules, this personalized module implements the algorithm, and gives the experimental demonstration, comparative experiments show that user's preferences model can better express the user's interest preferences and dynamic updates than the traditional vector space model and probabilistic model, can effectively improve accuracy of document retrieval. Therefore, based on the research of the search techniques of user's preferences, can significantly enhance the user's personalized experience, be able to more accurately and faster find target data for users.