The design of MOO agents: Implications from a study on multi-modal collaborative problem solving.

P. Dillenbourg1, P. Jermann1, D. Schneider1, D. Traum2, C. Buiu3

1 TECFA, School of Education and Psychology, University of Geneva, Route de Drize 9, 1227 Carouge Switzerland. Phone +41 (22) 705.96.93

2 UMIACS, A. V. Williams Building 9, University of Maryland, College Park, MD 20742 USA Phone: +1 (301) 405-113

3 Department of Computer Science and Control, University of Bucharest, Spl. Independentei 313 77206 Bucharest, Romania

Address for correspondence: Pierre.Dillenbourg@tecfa.unige.ch

Abstract. This contribution reports the results of an empirical study on computer-supported collaborative problem. Twenty pairs of subjects had to play a mystery solving game in a text-based virtual environment (TecfaMOO). This environment was enriched with a shared whiteboard. We analyzed how subjects use different communication tools to build common grounds. We present seven main observations which could inform the design of artificial MOO agents. They describe the extra-amount of reasoning required to participate into interactions and to build shared understanding. We propose a new type of artificial agent, observers, which do not solve the problem but compute statistics regarding interactions, which can be useful for human or artificial tutors or which could be displayed to the users themselves.


The design of MOO agents: Implications from a study on multi-modal collaborative problem solving - 21 MARCH 1997

Generated with Harlequin WebMaker