The effect of trust and team personality on multiagent teamwork performance

Abstract

Composed services consist of interacting services. Generally each service in a composed service is brought out by a different service provider. The quality of the composed service depends not only on the individual capabilities of the providers but also on how well they work together. In the pursuit of establishing effective teams, researchers propose several cognitive factors such as personality, trust, and leadership to model teams. In this theses, we study two significant factors: trust model and team personality. Existing trust models are geared towards identifying single services rather than composed services. However, in many settings it is important to find a group of service providers that can be trusted for a composed service. To address this, we propose a trust model that captures how trustworthy a group of service providers is for a particular composed service. The approach is based on capturing relations between services. Our proposed approach is tested on a modified version of ART Testbed. We compare our proposed model with an existing approach in the literature and show that capturing relations between services pays off infinding useful groups of service providers. For the second factor, we investigate the relationship between the team personality and teamwork performance. A promising personality model is composed of what is called Big Five Personality Traits: Agreeableness, Conscientiousness, Emotional stability, Extraversion, and Openness. We experimentally study the effect of these traits on multiagent teamwork. To do so, we model these traits and implement them in agents that can participate in ART Testbed by including interdependency attributes of teamwork. In this setup, we specifically study which traits are more significant than others for better performing teams, whether more trusted teams actually achieve a higher success rate than others, and whether heterogeneous teams perform better than homogeneous teams.