摘要： 将约束引入到可能性XML时，因分布节点值的不确定性，XML文档无法检验对涉及分布节点的约束的合法性。在一些场景中，要返回给用户符合约束的查询结果和概率，通常没有考虑去除不可能域分布来做条件计算，得出的并不是准确概率。为此，提出一种可能性XML的概念，给出可能性XML中可能域和约束的有效表达，通过提出的方案解决条件计算和准确概率计算的问题。实验结果表明，该算法的效率较高，条件计算后的结果更能被用户接受。

关键词: 可能性XML, 约束, 可能域描述符, 可能域集树, 条件计算

Abstract: While incorporating constraints in probabilistic XML, it becomes impossible to check the validity of the document against the constraints which involve distributional nodes, for the values of distributional nodes are uncertain. In some scenario, users should return the query answers which meet the constraints and their corresponding confidences. However, the confidences are not exact, for they are conditioned with respect to impossible world space. This paper introduces the concept of probabilistic XML, and gives an effective way to represent possible worlds and constraints in probabilistic XML, presents efficient techniques to solve both conditioning and exacts confidence computation problems. Experimental evaluation shows that the algorithm can work efficiently and the conditioned query answers are more acceptable to users.

Key words: probabilistic XML, constraint, possible World-set Descriptor(WSD), possible World-set Tree(WST), conditioning

中图分类号: 

• TP311