Using Genetic Algorithms to Manage Metric and Symbolic Temporal Information

M. Mouhoub (Canada)


Genetic Algorithms, Temporal Reasoning, Constraint Sat isfaction.


The aim of this work is to study the applicability of Genetic Algorithms (GAs) to handle both the numeric and symbolic aspects of time. Indeed, in applications such as planning, scheduling, computational linguistics and com putational models for molecular biology, managing qual itative and quantitative temporal information becomes an important issue. For instance, in scheduling applications we may have to look for a sequence of actions that should respect some precedence or other relative constraints be tween actions in addition to temporal windows restricting the time interval during which each action should occur. To manage temporal information using genetic algo rithms, we will present a method that converts a list of time information into a binary constraint satisfaction prob lem (CSP). Genetic algorithms are then used as an approx imation search method to check for the consistency of the CSP. Experimental comparison with other approximation methods based on local search demonstrates the efficiency of our method to deal with large size problems involving numeric and symbolic time information.

Important Links:

Go Back