The Agent–Object-Relationship metamodel: towards a unified view of state and behavior

摘要

We present an agent-oriented approach to the conceptual modeling of organizations and organizational information systems, called Agent–Object-Relationship (AOR) modeling, where an entity is either an agent, an event, an action, a claim, a commitment, or an ordinary object, and where special relationships between agents and events, actions, claims and commitments supplement the fundamental association, aggregation/composition and generalization relationship types of Entity-Relationship (ER) and Unified Modeling Language (UML) class modeling. Business processes are viewed as social interaction processes emerging from the behavior of the participating agents. In the proposed approach, behavior is primarily modeled by means of interaction patterns expressed in the form of reaction rules that are visualized in interaction pattern diagrams. It is an option, though, to use UML activity and statemachine diagrams, in addition.We propose an elaborate conceptual framework for agent-oriented modeling that is based on a set of 19 ontological principles including those of ER modeling, and a corresponding diagram language. In this approach, an organization is viewed as an institutional agent defining the rights and duties of its internal agents that act on behalf of it, and being involved in a number of interactions with external (and internal) agents, while an organizational information system is viewed as an artificial internal agent possessing a global view of the organization and interacting both with internal and with external agents on behalf of the organization.We argue that AOR modeling offers a research perspective to conceptually integrate the static, dynamic and deontic aspects of organizations and organizational information systems.