Agents supporting humans and organizations in open, dynamic environments

• Authors: Katia Sycara
• Venue: Proceedings of the first international joint conference on Autonomous agents and multiagent systems: part 3
• Year: 2002

The presence of the Digital Infosphere and the continuous growth of e-commerce have generated important shifts in the ways people and organizations get information and make decisions. These shifts necessitate increased automation and creation of infrastructure, standards and policies to enable machines to automatically access information, understand it, fuse it as needed and engage in collaborative problem solving to support decision making. Fulfilling such goals presents many challenges, including semantic interoperability, agent-based collaboration, information customization, automated and flexible service discovery and transaction across the Internet. Today, services are discovered and invoked manually by human users. In the near future, such service discovery and use will be mediated by agents acting on behalf of the humans" behalf. This opens the possibilities for agents and humans to be team partners and coordinate sharing information, responsibility and control according to the task requirements. There are many challenges to accomplish such collaboration. A crucial one is making the Web agent-understandable, i.e. allowing for semantic annotation of content. The combination of the Semantic Web and Agent Technology is the harbinger of the next Web revolution. Instead of being populated only with human-readable documents, the Web will be populated with agent-mediated services. In addition, agents will support human decision-making and human institutions through autonomous interactions, such as negotiations, coalition formation, and agent-mediated markets.In my Laboratory of Advanced Agent Technology at Carnegie Mellon University, we have been developing multiagent infrastructure, tools, and algorithms that comprise a Reusable Environment of Task-Structured Intelligent Networked Agents (RETSINA). This infrastructure can be used for developing distributed heterogeneous intelligent agents that interact in various ways including peer-to-peer manner, as well as agent-mediated services that describe themselves in semantically meaningful ways, discover one another dynamically, interoperate and compose themselves on-the-fly and on-demand, given particular tasks and goals to be fulfilled. This infrastructure has been used to support humans and organizations in open and dynamic environments, where information sources, agents and communication links may appear and disappear dynamically. The developed multiagent applications range from financial portfolio management, to distributed crisis action planning, team coordination, reactive and anticipatory assistance, location-based collaboration and e-commerce. In this talk, I will give an overview of our agent research, present current research results and future challenges. Up until now, this vision has been conceived and pursued mainly in academia and research labs. However, recent industrial interest in flexible interoperable automated transactions, Web Services, and the availability of tools to enable some form of service automation (e.g. UDDI, WSDL, X-lang, WSFL, e-speak, .NET, etc.) holds the promise of fast progress in this area.