Agents are autonomous software components that work with one another in a decentralized fashion to achieve some end. Agent systems have been used in Technology Enhanced Learning (TEL) before, but these applications seldom take advantage of the fact that each agent may have its own goals and strategies, which makes agent systems an attractive way of providing personalized learning. In particular, since agents can solve problems in a decentralized way, this makes them an attractive way of supporting informal learning. In this paper we use scenarios to examine how common problem solving techniques from the agents world (voting, coalition formation and auction systems) map to significant challenges for personalized and informal learning in the TEL world. Through an agent simulation we then show how an agent system might perform in one of those scenarios and explore how different agent strategies might influence the outcome. Based on this work we argue that agent systems provide a way of providing ultra-personalization of the learning process in a decentralized way and highlight equitability and scrutability as two key challenges for future investigation.
展开▼
