Artificial Intelligence
The consistent labeling problem and its algorithm: toward exact-case complexities and theory-based heuristics
Logical foundations of artificial intelligence
Logical foundations of artificial intelligence
POPL '87 Proceedings of the 14th ACM SIGACT-SIGPLAN symposium on Principles of programming languages
Constraint satisfaction problems in logic programming
ACM SIGART Bulletin
Constraint satisfaction algorithms
Computational Intelligence
Representation Selection for Constraint Satisfaction: A Case Study Using n-Queens
IEEE Expert: Intelligent Systems and Their Applications
Theory of Relational Databases
Theory of Relational Databases
Efficient processing of interactive relational data base queries expressed in logic
VLDB '81 Proceedings of the seventh international conference on Very Large Data Bases - Volume 7
A theoretical framework for consistency techniques in logic programming
IJCAI'87 Proceedings of the 10th international joint conference on Artificial intelligence - Volume 1
Hi-index | 0.00 |
We obtain here the complexity of solving a type of Prolog problem which Genesereth and Nilsson have called sequential constraint satisfaction. Such problems are of direct relevance to relational database retrieval as well as providing a tractable first step in analyzing Prolog problem-solving in the general case. The present paper provides the first analytic expressions for the expected complexity of solving sequential constraint satisfaction problems. These expressions provide a basis for the formal derivation of heuristics for such problems, analogous to the theory-based heuristics obtained by the author for traditional constraint satisfaction problem-solving. A first application has been in providing a formal basis for Warren's heuristic for optimally ordering the goals in a conjunctive query. Due to the incorporation of "constraint looseness" into the analysis, the expected complexity obtained here has the useful property that it is usually quite accurate even for individual problem instances, rather than only for the assumed underlying problem class as a whole. Heuristics based on these results can be expected to be equally instance-specific. Preliminary results for Warren's heuristic have shown this to be the case.