中文摘要：

为准确地推断可扩展标记语言（xML）关键字检索中的用户查询目标，提出一种目标节点推断方法。在获取目标节点时，考虑相应类型下XML节点出现的频率，以及用户输入关键字所在的不同位置对目标节点类型的影响，赋予词频不同的权重参数，同时引入XReal中XML文档树的层次信息，进行目标节点的推断。实验结果表明，该方法可以得到更准确的目标节点，提高查询准确率。

英文摘要：

In order to obtain user＇s query intent more accurately, this paper proposes a method to infer target node. It considers frequency of Extensible Markup Language（XML） node in the node type, and the effect when the same word appears at different positions, gives different weights to them. It keeps the level factor of XML document tree. Experimental results verify that the method is more efficiently to infer user＇s intention and improves the accuracy of the search.