Bayesian Network-Based Probabilistic XML Keywords Filtering

Chenjing Zhang, Kun Yue, Jinghua Zhu, Xiaoling Wang, Aoying Zhou

Abstract: Data uncertainty appears in many important XML applications. Recent probabilistic XML models represent different dependency correlations of sibling nodes by adding various kinds of distributional nodes, while there does not exist a uniform probability calculation method for different dependency correlations. Since Bayesian Networks can denote various dependency correlations among nodes just by conditional probability table(CPT), this paper proposes the Bayesian Networks based probabilistic XML model PrXML-BN, and combines SLCA semantic meaning of keyword query into Bayesian Networks, then implements keywords filtering on SLCA semantic meaning. To optimize the performance of keywords filtering, two optimization strategies are proposed in this paper. In the end, experiments verify the performance of keywords filtering algorithm based on SLCA in model PrXML-BN.