Abstract

Reasoning and planning at different levels of abstraction is an important skill in the game of Go, for both human and computer players. Over the years, adversarial planning approaches have become increasingly popular for dealing with large search spaces of two player games such as Go. This article describes a different approach to adversarial planning, based on an analysis of human Go playing. Abstract goals are decomposed by searching the space of lower level features. Goal checking is done using an abstraction operator.