On the Connections between Relational and XML Probabilistic Data Models

Abstract

A number of uncertain data models have been proposed, based on the notion of compact representations of probability distributions over possible worlds. In probabilistic relational models, tuples are annotated with probabilities or formulae over Boolean random variables. In probabilistic XML models, XML trees are augmented with nodes that specify probability distributions over their children. Both kinds of models have been extensively studied, with respect to their expressive power, compactness, and query efficiency, among other things. Probabilistic database systems have also been implemented, in both relational and XML settings. However, these studies have mostly been carried out independently and the translations between relational and XML models, as well as the impact for probabilistic relational databases of results about query complexity in probabilistic XML and vice versa, have not been made explicit: we detail such translations in this article, in both directions, study their impact in terms of complexity results, and present interesting open issues about the connections between relational and XML probabilistic data models.