A rule-based language for probabilistic reasoning

This paper suggests a rule-based logical language for probabilistic reasoning. This language is more expressive than the traditional propositional logic. It is particularly suited for reasoning with belief networks as its semantics explicitly captures the inherent conditional independencies. One salient feature of the proposed language is that it can be extended to include functional symbols and recursion. Moreover, our language subsumes the traditional rule-based language in relational databases.