Abstract: The research of semantic information retrieval method,improve the efficiency of retrieval.Differentiated information due to the characteristics of the difference is too big,the larger rejection in the retrieval process,the traditional semantic information retrieval model in view of the large differences in the process of information retrieval,is given priority to with testing rounds,efficiency is very low.For this,put forward a kind of semantic information retrieval method based on decision tree algorithm.Calculate according to the related theory,multi-level analytic fusion window function,and based on window function for different levels of data fusion,the differences in information fusion results are obtained.Based on the above results,the establishment of a decision tree,realization of semantic information retrieval.Experimental results show that the improved algorithm of semantic information retrieval,and can improve the efficiency of retrieval.