The importance of context-dependent learning in negotiation agents

DAN KROHLING; FEDERICO HERNÁNDEZ; OMAR CHIOTTI; ERNESTO MARTINEZ

Buenos Aires

Congreso; XIX Simposio Argentino de Inteligencia Artificial (ASAI) - JAIIO 47; 2018

Automated negotiation between artificial agents is essentialto deploy Cognitive Computing and Internet of Things. The behaviorof a negotiating agent depends significantly on the influence of environmental conditions or contextual variables, since they affect not only agiven agent preferences and strategies, but also those of other agents.Despite this, the existing literature on automated negotiation is scarceabout how to properly account for the effect of context-relevant variablesin learning and evolving strategies. In this paper, a novel context-drivenrepresentation for automated negotiation is proposed. Also, a simple negotiating agent that queries available information from its environment,internally models contextual variables, and learns how to take advantageof this knowledge by playing against himself using reinforcement learningis proposed. Through a set of episodes against negotiating agents in theexisting literature, it is shown that it makes no sense to negotiate without taking context-relevant variables into account. The context-awarenegotiating agent has been implemented in the GENIUS negotiation environment, and results obtained are significant and revealing.