Finding appropriate settings for fairness and engagement in a newly designed game through self-playing AI program: A case study using Japanese crossword game 'MyoGo Renju’

Abstract

This paper explores an innovative way to find the comfortable settings of a newly designed game under development using a computer program. A Japanese crossword game ‘MyoGo Renju’ has been chosen as a benchmark for this research, whereas some important aspects such as fairness and engagement are evaluated to find the most optimum settings for the player. The game ‘MyoGo Renju’ can be played by Japanese language learners with competitive purposes as well as educational purposes. An artificial intelligence program of the ‘MyoGo Renju’ has been developed and various parameters had been evaluated such as board size, word length, bonus point rule, block system, weighted score system, and round mode. The experiment is performed using self-playing Myogo AI where some interesting results have been demonstrated where the 5 × 5 board with the minimum word length of 3-g, 3 number of blocks, and 15 Hiragana characters chosen in a single round is expected to provide the best-expected fairness and engagement to the overall game experiences. The limitations and future works of the research are also discussed.