Abstract: Query recommendation is an important tool for improving searching efficiency. But traditional recommendation method cannot provide good recommendations for exploratory search. A novel model based on search behaviors for the above problem was proposed，which successfully build the exploratory search process by tentative query refinement and deterministic query refinement. Then a query recommendation algorithm based on the three factors affecting exploratory search process was proposed. Finally the results of comparative experiments showed that this model and algorithm achieved a good effect.

Key words: query recommendation, exploratory search, query chain, tentative query refinement, deterministic query refinement