Search recommendation model based on user search behavior and gradual forgetting collaborative filtering strategy  

作　　者：LIU Chuan-chang State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China 

出　　处：《The Journal of China Universities of Posts and Telecommunications》2010年第3期110-117,共8页中国邮电高校学报（英文版）

基　　金：supported by the National Natural Science Foundation of China (60432010);the National Basic Research Program of China (2007CB307103);the Fundamental Research Funds for the Central Universities (2009RC0507);Important Science & Technology Specific Project of Guizhou Province (【2007】6017)

摘　　要：The existing search engines are lack of the consideration of personalization and display the same search results for different users despite their differences in interesting and purpose. By analyzing user＇s dynamic search behavior, the paper introduces a new method of using a keyword query graph to express user＇s dynamic search behavior, and uses Bayesian network to construct the prior probability of keyword selection and the migration probability between keywords for each user. To reflect the dynamic changes of the user＇s preference, the paper introduces non-lineal gradual forgetting collaborative filtering strategy into the personalized search recommendation model. By calculating the similarity between each two users, the model can do the recommendation based on neighbors and be used to construct the personalized search engine.The existing search engines are lack of the consideration of personalization and display the same search results for different users despite their differences in interesting and purpose. By analyzing user＇s dynamic search behavior, the paper introduces a new method of using a keyword query graph to express user＇s dynamic search behavior, and uses Bayesian network to construct the prior probability of keyword selection and the migration probability between keywords for each user. To reflect the dynamic changes of the user＇s preference, the paper introduces non-lineal gradual forgetting collaborative filtering strategy into the personalized search recommendation model. By calculating the similarity between each two users, the model can do the recommendation based on neighbors and be used to construct the personalized search engine.

关 键 词：search recommendation model search behavior expression keyword query graph gradual forgetting collaborative filtering 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]