Abstract: A keywords retrieval method of semantic relevant was proposed to deal with the loss of semantics information in XML keywords retrieval. The implied semantics in document were fetched by using the semi-structured feature of XML document; the user inquiry intents were also captured by analyzing the inquiry syntax. And then, the elements satisfying the demands were retrieved according to user inquiry intent. Finally, in combination with semantics of the document, the expressions of inquiry results were improved by using the semantic relevant entity sub-tree set, instead of the traditional Smallest Lowest Common Ancestor (SLCA). The experimental results indicate that the precision ratio of keywords retrieval can be improved by using this method.

Key words: smallest lowest common ancestor, keyword search, semantic relevant entity sub tree set, precision ratio

摘要： 为了解决XML关键字查询中语义信息丢失的问题,提出了一种语义相关的关键字检索方法。利用文档的半结构化特点提取文档隐含的语义,利用查询语法捕获用户查询意图,然后根据用户意图查询满足条件的元素,并结合文档语义,由最小最近公共祖先改进为语义相关实体子树集来表达查询结果。实验结果表明,该方法能够有效提高关键字检索结果的查准率。

关键词: 最小最近公共祖先, 关键字查询, 语义相关实体子树集, 查准率