This paper describes a new approach to design modern information systems that offer an integrated access to the data and knowledge that is available in local applications. By integrating the local data management activities into one transparent information distribution process, modern organizations can offer better support for its workers in the execution and coordination of their work activities. Observing practical applications of decentralized, autonomous and heterogeneous information systems we see deficiencies in currently available approaches to model such information systems. They do not acknowledge the pivotal role that (informal) communicating workers play in the context of an entire organization. The interaction between the members of social and informal groups of employees makes it possible that (in many of today’s information-intensive enterprises) the local structured procedures can be effectively and flexibly integrated into global work processes supporting the business goals. Traditional design techniques concentrate on either the structured local procedures (and its local database applications), the structured global process (and its global business goals), or the informal (less structured) communication between individuals. We suggest to combine an activity-based model (suited to describe the structured parts of the processes) with a goal- or conversation-based model to tie the different elements together. Using an agent architecture we show that it is possible to implement this integrated approach. The different types of cooperating agents support the individual workers by assessing the goal of the activity, the applicability of the standard procedure, and the availability of alternative knowledge and information in order to supply the necessary information.
A proof is whatever convinces me. Shimon Even (1935–2004) The glory attached to the creativity involved in finding proofs makes us forget that it is the less glorified process of verification that gives proofs their value. Conceptually speaking, proofs are secondary to the verification process, whereas technically speaking, proof systems are defined in terms of their verification procedures. The notion of a verification procedure presumes the notion of computation and furthermore the notion of efficient computation. This implicit stipulation is made explicit in the definition of NP , where efficient computation is associated with deterministic polynomial-time algorithms. However, as argued next, we can gain a lot if we are willing to take a somewhat non-traditional step and allow probabilistic verification procedures. In this chapter, we shall study three types of probabilistic proof systems, called interactive proofs, zero-knowledge proofs , and probabilistic checkable proofs . In each of these three cases, we shall present fascinating results that cannot be obtained when considering the analogous deterministic proof systems. Summary: The association of efficient procedures with deterministic polynomial-time procedures is the basis for viewing NP-proof systems as the canonical formulation of proof systems (with efficient verification procedures). Allowing probabilistic verification procedures and, moreover, ruling by statistical evidence gives rise to various types of probabilistic proof systems. Indeed, these probabilistic proof systems carry a probability of error (which is explicitly bounded and can be reduced by successive applications of the proof system), yet they offer various advantages over the traditional (deterministic and errorless) proof systems. […]