Bayesian Networks Model for Xml Documents Ranking

As more and more data are described, stored, exchanged and represented by XML, the abilities of information retrieval for XML documents become increasingly important. However, the retrieval results to users are quite large. To text-rich XML documents´ retrieval, a structured index method is designed at first, which accounts for the structure and content of each document. Then each XML document is modeled through a Bayesian network to handle both structure and content for the document. This paper also presents the inference process for computing the probability of each document on the given query. Finally documents are ranked according to the probabilities in descent. The experiments indicate that this framework can reduce the workloads and the complications, and also improve the recall and precision