The Wolf in Sheep’s Clothing: How Locks Can Gently Control Collaboration

Abstract

A typical Cooperative Information System comprises a number of heterogeneous, autonomous, and distributed component software systems, like (multi)database systems, knowledge-based systems (especially expert systems) or multimedia systems. Each such system can be seen as an (information) agent which is capable of providing information and/or solving problems on its domain of expertise. In order to be able to solve a broad spectrum of complex, information-intensive problems, agents interact in different ways, often in an ad hoc manner; i.e. agents are capable of treating problems which can only be solved by selectively and individually fusing and combining problem-solving expertise and data/information sources from the various agents in the network. Such tasks may be executed autonomously, collaboratively or cooperatively, depending on the complexity and nature of the task. In general, if complex problems are to be solved they are subdivided into less complex subtasks which again are subdivided into subtasks and so on. In a similar way agents are sometimes arranged in higher-level coalitions or clusters, each coalition being grouped around some common domain of expertise. Coalitions can overlap in that agents can belong to more than one coalition (Haghjoo et al. 1993). This usually will result in a more cooperative style of work within such a coalition and a more isolated style of work between them.