Designing Knowledge-Based Rule-Based Agents in Schnapsen: Leveraging Human Winning Strategies

Yi Zhou, Xiaowu Zhao, Dou Zheng

Abstract: This research focuses on developing a rule-based computer bot for the Schnapsen game, a trick-taking card game known for its strategic depth. In contrast to well-studied board games like chess and Go, Schnapsen poses unique challenges due to incomplete information and the influence of chance events. Leveraging insights from human strategies, we construct a rule-based system for the bot, with a particular focus on moderately aggressive gameplay. The experiment setup involves measuring the bot’s performance against random, rdeep, and ml bots, with a sample of 1000 plays for each opponent. The results showcase the bot’s competitive winning rates, providing insights into the efficacy of rule-based strategies in the context of Schnapsen.