Trust-Based Agent Evaluation in Collaborative Systems.

Agent evaluation is an important topic in collaborative systems. Agent evaluation is to accurately assess the ability of the agent to complete a task. The ability of the agent is demonstrated by the result of the task, and the ability changes over time. Thus, this paper studies the evaluation of agent in collaborative systems. Considering the independent and interaction abilities of an agent for a role, as well as the historical and current evaluation value, we propose a trust-based evaluation approach that promotes the effectiveness of evaluation. The experimental results show that this approach can obtain more accurate results when the agent capability changes fast and there are only a few known evaluation values.