(52) Go Playing Game System "GOG"(The Sequential Version)
Machine: PSI
Environment: SIMPOS
Language: ESP
Source Code: 1.5 MB
Documents: Manual (English/Japanese)
Overview
Go Playing Game System.
Configuration
┌───── Playing System (Board Recognition
│ Move Generation, Local Area Search)
GOG ────┼───── Knowledge Editor
└───── Evaluation Tool
Function
The "GOG" plays whole Go game by the Chinese rule. We can edit some
"candidate knowledges" using Knowledge editor. We can also evaluate
the improved candidate knowledges using Evaluation tool.
Feature
The approach of GOG is based on simulating a human player's thinking
mechanism. As a Go-playing program requires basic AI techniques such
as searching, processing ambiguous patterns, exceptional processing,
and cooperative problem solving, it is a suitable research subject for
knowledge processing technologies.
Due to the high branching factor of the Go game tree, computer chess
type full board search becomes hopeless in this domain. In the stage
of board recognition and planning, GOG simulates the human player's
"focus of attention".
References
- Methodology of GOSEDAI, N.Sanechika et al., ICOT-TR 717, The
Specifications of "GO Generation", N.Sanechika et al., ICOT-TR 720
FTP
- Go Playing Game System "GOG"(The Sequential Version) [712K]
www-admin@icot.or.jp