Abstract: In this paper, a Bayesian model for XML document information retrieval was proposed, supporting content and structure queries. The topology and conditional probabilities of the Bayesian model were mined from the XML document collection by statistics of terms, elements and structural units contained in the collection. Combined with probability functions, the relevance of a document to a given structured query was obtained by an inference process through a complex dependences network of this model. Experimental results on the INEX corpus of XML documents show the validity and reliability.

Key words: Bayesian network, XML information retrieval, structured query, probability function

摘要： 提出了一种新的基于贝叶斯网络对XML文档信息进行查询的模型方法。该模型支持针对XML文档信息的结构化查询。基于XML信息查询的特点，利用XML数据集中语词、元素和结构化单元的统计信息对模型的拓扑结构和条件概率进行了学习；结合概率函数的方法，利用模型的概率推理进程对XML文档和结构化查询条件的相关度进行了估算。最后在基于INEX测试集的实验中证明了该方法的有效性和可靠性。

关键词: 贝叶斯网络, XML信息检索, 结构化查询, 概率函数