中文摘要：

为了进一步改进个性化搜索方法，通过对现有个性化搜索方法的研究，提出了一种新的搜索方法。该方法从用户兴趣相关性出发，将用户配置文件与传统个性化搜索相结合。在传统TF-IDF方法的基础上，提出了一种综合考虑标签总引用次数和配置文件中标签总数的新方法，用于获取用户配置文件与资源配置文件中的标签权重；设计了基于余弦相似性计算并综合匹配的标签个数的资源相关性计算方法。通过MovieLens数据集实验，验证了提出方法的有效性。

英文摘要：

In order to further improve personalized search methods, by researching existing personalized search methods, this paper proposed a new search method. The method started from the user＇ s interest correlation and combined the user profile with the traditional personalized search. Based on the traditional TF-IDF method, it put forward a new comprehensive method which took the total number of tags in the configuration flies and the total citations of a tag into consideration to get the weight of the tags in the user profile and resource profile. Then it designed a correlation calculation method of resources, which based on the cosine similarity calculation and took the number of matching tags into consideration. Experimental results with Mov- ieLens data set verify the validity of the proposed method.