Personalized information retrieval base-on user interest state

Different users have different needs, even the same user may have different desires in different time, Personalized Information Retrieval makes search results meet different users´ information requirement. In this paper, a kind of user interest recognition algorithm is proposed, which can analyze the user interest state and identify the user´s Temporary interest; And a state-based user interest model is developed, In this model, user interest is recognized by the algorithm mentioned above, and users´ characteristic is extracted and Dynamic weighted, then gray relational analysis is introduced to for the comprehensive consideration of two aspects above. The experimental result indicates that the average push accuracy is above 70% and the push service is more accurate for the user who has a long interest cycle.