Abstract: In order to effectively solve the data correction,elimination and integration problems for Deep Web database,make the Deep Web database entity query performance with optimization,the database entity recognition intelligent semantic query method based on autocorrelation characteristics was proposed.The query model consisted of text matching model,semantic correlation characteristic analysis model and the statistical model,and semantic correlation feature extraction algorithm was proposed.The semantic information constraint rule was defined for optimal approximation database entity recognition results.The semantic correlation characteristics were used to establish the surface correlation knowledge base,the correction,elimination,integration was realized in the data query.The mathematical derivation proves that the stability of the algorithm,and simulation results show that algorithm can consider,text feature,semantic correlation feature and constraint rules,the accuracy and completeness of data query increased significantly and effectively.