Abstract

Stochastic Logic Programs (SLPs) are a generalisation of Hidden Markov Models (HMMs), stochastic context-free grammars, and undirected Bayes' nets. A pure stochastic logic program consists of a set of labelled clauses p:C where p is in the interval [0; 1] and C is a first-order range-restricted definite clause. SLPs have applications both as mechanisms for efficiently sampling from a given distribution and as representations for probabilistic knowledge. This paper discusses a distributional semantics for SLPs. Additionally two derivational mechanisms are introduced and analysed. These are 1) stochastic SLD (Selection-function-Linear-resolutionfor -Definite-clauses) refutation and 2) a program for enumerating the least Herbrand distributional model of an SLP in descending order. It is shown that as with logic programs unfolding transformations preserve the models of an SLP. Lastly, we discuss how Bayes' Theorem might be used to devise algorithms for learning SLPs. 1...

Citations