With the help of a model for discrete optimization problems, we show that a large number of heuristic search procedures (for searching state-space graphs, AND/OR graphs, game trees, etc. ) of artificial intelligence (AI), and dynamic programming (DP) and branch-and-bound (B&B) procedures of operations research use problem-specific knowledge in a framework based upon context-free grammar. The model reveals the true nature of these procedures, and aids in synthesizing new variations as well as generalizations and parallel implementations of these procedures. The paper concludes by commenting upon how this model may be generalized and made more powerful to encompass a greater variety of problems, and to help synthesize more efficient search procedures.
|Original language||English (US)|
|Title of host publication||Proceedings of the Annual Conference of the Association for Computing Machinery|
|Publisher||ACM (Order n 401840)|
|Number of pages||6|
|State||Published - Dec 1 1984|