A retrieve algorithm that incorporates the semantic information of the words into traditional retrieve function of Lucene is proposed. The proposed method improves the important components of existing retrieve similarity functions with semantic information

and selects the appropriate measure of semantic similarity to compute the semantic similarity between the query words and text corpus by using the external dictionary Wordnet. With the semantic similarity

the algorithm implements semantic information retrieve and can sort the retrieved text documents according to the semantic similarity between query words and text documents. The experimental results show that the proposed method can improve the precision of document retrieval effectively.