Behavioral semantics of nonrecursive control structures
Programming Symposium, Proceedings Colloque sur la Programmation
The control structure facilities of ECL
Proceedings of the international symposium on Extensible languages
Some transformations for developing recursive programs
Proceedings of the international conference on Reliable software
POPL '76 Proceedings of the 3rd ACM SIGACT-SIGPLAN symposium on Principles on programming languages
Hi-index | 0.00 |
A POP-2 package known as AL1 ("Advice Language 1") has been developed during the Spring semester by the CS397DM graduate class while the author was visiting the University of Illinois. The object was to facilitate the transfer of specialist knowledge about chess end-games into machine memory. The package comprises two main modules:An Advice module (input: a board-state; output: an advice-list). This module is partitioned into "Advice Tables" corresponding to a subdivision of the task domain into sub-domains.A Search module (inputs: a board-state and an advice-list; output: a "forcing-tree strategy" for securing specified goals). This module includes move-generation, the only part of the package which is specific to chess.For executing strategies, a tree-lookup routine generates play against an opponent. A Table-editor allows the user to display, modify, or extend the Tables. These are of two kinds: A single Master Table decides, on the basis of properties of the input board-state, to which of a "committee of experts" it should be referred. The "experts" themselves are the individual Advice Tables, each written for a specific sub-domain. A fully tested Table for the elementary sub-domain King and Rook versus King (elementary to play, not to program) showed an order-of-magnitude advantage over a conventional programming approach.