(22) A Programming System for Statistical Modeling

We have started PRISM (PRogramming In Statistical Modeling) project to explore the idea of logic programs that interact with the real world, learn from data and change adaptively their behaviors. Theoretically it originated from the combination of fixed point semantics and probability theory, but was spurred by the recent success of symbolic statistical model such as hidden markov model and Bayesian networks. The project aims to establish a common basis for those diverse activities by providing a descriptive means for logical-statistical relationships together with their learning algorithms in a single framework.