Demo:

A Demonstration of the Agent Reputation and Trust (ART) Testbed for Experimentation and Competition
Karen K. Fullam, Tomas B. Klos, Guillaume Muller, Jordi Sabater, Zvi Topol, K. Suzanne Barber, Jeffrey S. Rosenschein, and Laurent Vercouter


This demonstration will detail the Agent Reputation and Trust (ART) Testbed, a tool through which agent trust researchers can 1) perform easily-repeatable experiments in a common environment and 2) participate in competitions with other researchers to identify the most promising trust technologies. The demonstration will be presented in a software live demonstration format, supplemented by video and powerpoint slides. A related technical paper has been accepted for presentation at the AAMAS general conference.

The Testbed presents researchers with an art appraisal domain problem: agents function as painting appraisers with varying levels of expertise in different artistic eras. If an appraiser does not have the expertise to complete the appraisal, it can purchase (possibly false or inaccurate) opinions from other appraisers. Appraisers receive more clients, and thus more profit, for producing more accurate appraisals. Appraisers can also purchase reputation information from each other, which helps appraisers know from which other appraisers to request opinions.

As a versatile, universal experimentation site, the Testbed fosters a cohesive scoping of trust research problems; researchers are united toward a common challenge, out of which can come solutions to these problems via unified experimentation methods. Through objective, well-defined metrics, the Testbed provides researchers with tools for comparing and validating their approaches. The Testbed also serves as an objective means of presenting technology features (both advantages and disadvantages) to the research community. In addition, the Testbed places trust research in the public spotlight, improving confidence in the technology and highlighting relevant applications.

This demonstration will 1) introduce researchers to the art appraisal problem domain and its competition rules, and 2) exhibit how researchers can navigate the Testbed user interfaces to initiate, join, and view games. The authors plan to release the Testbed software to the research community in July of 2005 for a competition planned in July of 2006.