关键词: 可扩展标示语言（XML）, 半结构文档, 相似度

Abstract: With more and more application of semi-structure data, the research of XML document similarity becomes essential in the database and information retrieval communities. Given set of XML documents D and the user query q, XML retrieval is to find out the XML documents from the D which satisfies q. In order to search efficiently, a new approach is presented to calculate similarity between two XML documents. The approach is divided into three steps. The user’s query q is expanded to q' by including the synonyms of q based on WordNet. q' and each XML document in D are allocated to digital signatures. After eliminating the irrelevant documents in D according to the signatures matching, a subset D' of  D is got. Precise matching between q and D' is presented and final results are got.

Key words: Extensive Makeup Language（XML）, semi-structure data, similarity