Trust in agent-based software

Many modern computer applications are open distributed systems in which the (very many) constituent components are spread throughout a network, in a decentralized control regime, and which are subject to constant change throughout the system’s lifetime. Examples include the grid (Foster and Kesselman 1998), peer-to-peer computing (Gong 2002), the semantic web (BernersLee et al. 2001), web services (Cerami 2002), e-business (Deitel et al. 2001), m-commerce (Sadeh 2002; Vulkan 1999), autonomic computing (Kephart and Chess 2003) and pervasive computing environments (Schmeck et al. 2002). In all of these cases, however, there is a need to have autonomous components that act and interact in flexible ways in order to achieve their design objectives in uncertain and dynamic environments (Simon 1996). Given this, agent-based computing has been advocated as the natural computation model for such systems (Jennings 2001).