Production of various strategies and position control for Monte-Carlo Go — Entertaining human players

Abstract

Thanks to the continued development of tree search algorithms, of more precise evaluation functions, and of faster hardware, computer Go and computer Shogi have now reached a level of strength sufficient for most amateur players. However, the research about entertaining and coaching human players of board games is still very limited. In this paper, we try first to define what are the requirements for entertaining human players in computer board games. Then, we describe the different approaches that we have experimented in the case of Monte-Carlo computer Go.