The importance of context-dependent learning in negotiation agents

Abstract

Automated negotiation between artificial agents is essential to deploy Cognitive Computing and Internet of Things. In this sense, the behavior of those negotiation agents depend significantly on the influence of environmental variables, facts, and events, which made up the context of the negotiation game. This context affects not only a given agent preferences and strategies, but also those of his opponents. In spite of this, the existing literature on automated negotiation is scarce about how to properly account for the effect of the context in learning and evolving strategies. In this paper, a novel context-driven representation of the negotiation game is introduced. Also, a simple negotiation agent that queries available information from context variables, internally models them, and learns how to take advantage of this knowledge by playing against himself using reinforcement learning is proposed. Through a set of episodes of our context-aware agent against other negotiation agents inthe existing literature, it is shown that it makes no sense to negotiate without taking relevant context variables into account. Our context-aware negotiation agent has been implemented in the GENIUS tool. Results obtained are significant and quite revealing about the role of self-play in learning to negotiate