This document describes computer go researches carried out since 1997 at the university of Paris 5 to obtain the accreditation to supervise research. Computer games have witnessed improvements performed in AI. However, go programming constitutes a very hard task for two reasons. First, the size of the game tree forbids any global tree search, and second, a good evaluation function is very hard to find out. The research presented within the Indigo project aims at producing a go program “as strong as possible ” while publishing the results. Between 1997 and 2002, the approach was based on knowledge, and had many links with many AI sub-domains. The scientific results are described for each sub-domain: DAI, cognitive science, fuzzy logic, uncertainty representation, combinatorial games, retrograde analysis, tree search, evaluation function and spatial reasoning. In 2002, the Monte Carlo approach was launched, and gave promising results. This fact brought about the programming of an approach associating Monte Carlo and knowledge in 2003. Then, that association yielded good results at the computer olympiads held in Graz in 2003. Thus, the Monte Carlo approach will be pursued in the future, and will be associated with bayesian and reinforcement learning.

Go, a popular game of strategy, is not amenable to the traditional AI techniques for game playing. Because of the game’s hierarchical and temporal nature, the use of structural representation is a logical approach to try. This thesis presents a preliminary representation of the game of Go in the structural language of the Evolving Transformation System (ETS). An ETS class hierarchy designed to capture progressively complex aspects of Go positions is presented. Individual Go stones are represented as elements of a few low-level classes, these classes are the constituent pieces of higher-level classes corresponding to blocks of stones, which are in turn assembled into classes corresponding to the evolution of larger shapes. The presented class hierarchy could form the basis of the first entirely class-based Goplayer, meaning that such a program’s move selection would be directed by generative class descriptions.