Skip to Main Content
Most recent document standards like XML urge Information Retrieval to design and implement systems to index, retrieve and present documents according to the given document structure. We present a Bayesian model for XML document retrieval. This model is based on Bayesian networks that enable to represent and quantify the relations between the structural components of the document and whose conditional probabilities are learnt from a labeled collection of structured documents - which is composed of documents, queries and their associated assessments. It allows us to focus on free SO (structure only) queries and obtain the relevance of a document to a given structural query by means of an inference process through a complex network of dependences. Some preliminary results on the system implementation are also presented.