Personalization recommendation (PR) is an important feature of future search engine. It is a valid method for lightening the user’s burden on information retrieval. This paper presents a new approach for PR based on user-focus. To construct user-focus for a user, a new algorithm WeightedFP for mining weighted frequent item set is given. The transactions for WeightedFP to be dealt with are the entire query requests of a user at a period of time and items in a transaction are non-noise words in the query request corresponding with the transaction. Each word as an item in itemset has a weight expressing the importance description of the word to the user. Experimental result shows that the implementation of PR based on user-focus can lighten the user’s burden caused by the work of filtering valid information from vast information in some extent while time requirement of TR is satisfied well.