Abstract: Current text search engines always have low search efficiency due to their keyword matching method.Based on the comparison of previous works,a thematic search strategy is proposed.The main idea of this strategy is grounded on the rich information implicated by the link structure of Wikipedia.It defines a measure of distance between words in terms of DBW,underpinned by computational thematic communities model.In this way,the authors can use this algorithm to rank and reorient the Key words to discover the closest keyword clusters and improve the quality of searching result.Introducing users' appraisal mechanism and making comparison with the traditional search engines' outcomes in experiment prove that the strategy expands the thematic coverage and maintains a high users' intent recognition at the same time.