Developing Strategies for the ART Domain

In this paper we propose the design of an agent for the ART Testbed, a tool created with the goal of objectively evaluate different trust strategies. The agent design includes a trust model and a strategy for decision making. The trust model is based on the three components of trust considered in ART, namely direct, indirect (reputation) and self trust (certainty). It also incorporates a variable time window size based on the available information that allows the agent to easily adapt to possible changes in the environment. The decision-making strategy uses the information provided by the trust model to take the best decisions to achieve the most benefits for the agent. This decision making tackles the exploration versus exploitation problem since the agent has to decide when to interact with the known agents and when to look for new ones. The agent, called Uno2008, competed in and won the Third International ART Testbed Competition held at AAMAS in March 2008.

• Type: Conference
• Year: 2009
• Where: CAEPIA
• Authors: Javier Murillo, Víctor Muñoz, Beatriz López, Dídac Busquets
• Keywords: Agent Design, Artificial Intelligence, CAEPIA 2009, Decision Making, Trust Model