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 s...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.展开更多
基金supported by the National Natural Science Foundation of China (60432010)the National Basic Research Program of China (2007CB307103)+1 种基金the Fundamental Research Funds for the Central Universities (2009RC0507)Important Science & Technology Specific Project of Guizhou Province (【2007】6017)
文摘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.
