Preference-Based Search and Multi-Criteria Optimization

Abstract

Many real-world AI problems (e.g. in configuration) are weakly constrained, thus requiring a mechanism for characterizing and finding the preferred solutions. Preference-based search (PBS) exploits preferences between decisions to focus search to preferred solutions, but does not efficiently treat preferences on defined criteria such as the total price or quality of a configuration. We generalize PBS to compute balanced, extreme, and Pareto-optimal solutions for general CSP's, thus handling preferences on and between multiple criteria. A master-PBS selects criteria based on trade-offs and preferences and passes them as optimization objective to a sub-PBS that performs a constraint-based Branch-and-Bound search. We project the preferences of the selected criterion to the search decisions to provide a search heuristics and to reduce search effort, thus giving the criterion a high impact on the search. The resulting method will particularly be effective for CSP's with large domains that arise if configuration catalogs are large.