2011 Fiscal Year Final Research Report
Description of Intelligent Games using Language and Mathematical Models
| Project/Area Number |
21500144
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Kyushu Institute of Technology |
|---|
Principal Investigator |
NAKAMURA Teigo 九州工業大学, 大学院・情報工学研究院, 准教授 (40198221)
|
|---|
| Project Period (FY) |
2009 – 2011
|
| Keywords | 知能情報処理 |
|---|
| Research Abstract |
We characterized the feature of moves from the following three aspects :(a) absolute board position of the move,(b) relative board position from the previous opponent' s move,(c) surrounding configuration of the move, and encoded the game records into text characters. We constructed an n-gram model from the texts and showed that the model can be used not only for move prediction but also for reducing the number of second best moves and discriminating moves for commentary. We analyzed some typical playing styles of human Go players and found some factors with significant difference between different playing styles.
|
