MibML: A conceptual modeling grammar for multiagent-based business integrative information systems.

摘要

With intense global competition and highly dynamic marketplaces, it is necessary for enterprises to achieve agility through integrated business information systems (IBIS). Recently, multi-agent systems (MAS) are becoming a key technology in developing IBIS applications in business domains. Although proliferation of agent applications boosts various approaches and architectures for solving the business and IS integration problem, most of these approaches have their roots in MAS and incorporate too many implementation-level considerations. There is currently a lack of a unifying framework which synthesizes characteristics of both IBIS and the MAS paradigm in the conceptual level. The goal of this study is to fill this void by developing a conceptual modeling grammar for multiagent-based integrative business information systems (MIBIS).; We first review the IBIS modeling and MAS literatures and confirm that the MAS paradigm provides an excellent approach for modeling and implementing IBIS systems. We synthesize these two bodies of literature and propose the MIBIS architecture and a set of fundamental constructs that are minimally required for MIBIS modeling. These constructs---Goal, Role, Interaction, Task, Information, Knowledge, and Agent---form the foundation of the conceptual modeling grammar MibML (Multiagent-based Integrative Business Modeling Language). The grammar is further enhanced as a formal model by defining highly refined constructs, definitions, relationships, and axioms in first order logic. The application of MibML is illustrated in a proof-of-concept example which models an online PC ordering system. In addition, the ontological expressiveness of the MibML grammar is evaluated by mapping the MibML constructs into upper-level ontological constructs in the Bunge-Wand-Weber (BWW) ontology. The comparison enables us to draw explicit rules which help map business domain knowledge into MibML constructs in MIBIS modeling.; This study contributes to both theory and practice in the realm of MIBIS. In theory, it lays the foundation of MIBIS modeling by providing a minimal set of orthogonal constructs with clear ontological semantics, each of them capturing a different and necessary perspective of MIBIS. In practice, it provides a tool to help system analysts construct MIBIS models more easily and in a consistent manner with minimum semantic gaps.