Catalogue of Artificial Intelligence Techniques
Categories: Problem Solving
Author(s): Austin Tate
Some systems do not have a fixed (Goal-driven or Data-driven) directional approach to solving a problem. Instead a current `focus' for the search is identified on the basis of the most constrained way forwards. This may be suggested by comparison of the current goals with the initial world model state, by consideration of the number of likely outcomes of making a selection, by the degree to which goals are instantiated, etc. Any problem-solving component may summarise its requirements for the solution as constraints on possible solutions or restrictions of the values of variables representing objects being manipulated. It can then suspend its operations until further information becomes available on which a more definite choice can be made. Many such systems operate with a Blackboard through which the various components can communicate via constraint information. The scheduling of the various tasks associated with arriving at a solution may also be dealt with through the blackboard.
- Stallman, R.M. and Sussman, G.J., Forward-reasoning and dependency-directed backtracking Artificial Intelligence 9 (1977), 135--196.