This paper describes an experiment to measure the playing of strength of Shogi players, man or machine, using a test set with 48 next-move problems. Each problem in the test set has been selected from games played by grandmasters, using some specific criteria. In Shogi there is no such rating system as the ELO system in chess to measure the playing strength of players and there was hitherto no test set for Shogi computers like the Bratko-Kopec test set for chess computers. In this paper, we first give a short sketch of test sets in use for chess computers, focusing on our interest in obtaining a similar test set for Shogi computers. Then our recent work on measuring the playing strength of Shogi computers using a newly devised test set is described, including some experiments performed with computer and human subjects.
|Title of host publication||CSC '96 Proceedings of the 1996 ACM 24th annual conference on Computer science|
|Publisher||ACM New York|
|Number of pages||8|
|Publication status||Published - 1996|
Matsubara, H., Iida, H., & Uiterwijk, J. W. H. M. (1996). A Shogi-Computer Test Set. In CSC '96 Proceedings of the 1996 ACM 24th annual conference on Computer science (pp. 139-146). ACM New York. http://www.mendeley.com/research/shogicomputer-test-set