Handling Probabilistic Uncertainty in Constraint Logic Programming

Abstract

The paper describes the approach and prototype of the tool for handling uncertainty in probabilistic interpretation in the framework of Constraint Logic Programming (CLP). The approach is intended to formulate real-life scheduling, planing, manufacturing problems more adequate to real situation with regard to uncertainty of input information. The main features and implementation of the tool integrated in the CLP(R) system [1] are described briefly. Theoretical background of the approach is a probabilistic logic introduced by N.Nilsson [2]. Calculation of probabilities for the goals in probabilistic logic programs is performed by means of Monte-Carlo method as suggested in [3]. The pilot application for dealing with network based project planing is presented in order to demonstrate the technology of the tool usage. We also discuss possible ways of speeding up the Logical Inference with Probability (LIP) by means of the so-called Success Formula that allows to skip unproductive trials and increase efficiency.