Most information systems development methodologies are based on conceptual modeling of static and dynamic views, which are represented by totally different types of diagrams. Understanding of the interplay among interactive, behavioral and structural aspects of specifications is necessary for identification of semantic integrity problems between business process and business data. Typically, semantic inconsistencies and discontinuities between collections of conceptual representations are not easy to detect and to comprehend for information system designers due to static and dynamic aspects of models being visualized in isolation. The goal of this paper is to present a modeling approach for semantic integration and evolution of static and dynamic aspects of conceptual models. Visualization of interplay among structural, interactive and behavioral aspects of computation-neutral representations helps to understand crosscutting concerns and integrity problems of information system conceptualizations. The main advantage of the presented conceptual modeling approach is stability and flexibility of diagrams in dealing with the evolutionary changes of requirements. Therefore, the developed modeling foundation is targeted to both business managers and information system designers for the purpose of computation-neutral integration and evolution of information systems specifications.
TopIntroduction
Information system architectures can be defined on various granularity levels and expressed by modeling views, which are normally represented by using different types of diagrams. It is common to the industrial versions of system analysis and design methods to distinguish disparate views and dimensions of enterprise architecture (Zachman, 1987). The Zachman framework can be viewed as taxonomy for understanding of different types of diagrams. It defines various dimensions of business application and data architectures such as What, How, Who, Where, When and Why. Integrity of different architecture views and dimensions is a fundamental problem in most Information System (IS) methodologies. Semiformal IS modeling methods such as structured analysis and design (Gane & Sarson, 1979; DeMarco, 1979; Yourdon & Constantine, 1979) are plagued by the paradigm mismatch between static and dynamic types of diagrams, which represent business processes and business data in isolation.
Object-oriented design (Booch et al., 1999) languages were introduced to overcome the integrity problems of static and dynamic aspects. Unified Modeling Language (UML) (OMG, 2010) provides various types of diagrams, which are applied for representation of interactive, behavioral and structural aspects of information system specifications. Every modeling approach, which covers a collection of different diagrams, must contain a systematic method for detection of inter-model inconsistency and incompleteness. The static and dynamic aspects of IS specifications are complimentary and they cannot be analyzed in isolation. The structural aspects describe characteristics of objects, which are invariant in time. The interactive and behavioral aspects describe dynamic characteristics of objects over time. All object-oriented design languages are based on collections of meta-models (Glinz, 2000). There is often semantic discontinuity and overlapping in IS specifications, because static and dynamic constructs of meta-models and notations do not fit perfectly. Semantic integrity of diagrams is difficult to achieve in this situation. This problem was addressed in the development of the ArchiMate language, which can be used for bridging from business to Information Technology (IT) layers of enterprise system architecture (Jonkers et al., 2004). A number of ontologically grounded modeling rules proposed by Evermann and Wand (2009) are necessary to achieve integrity among UML diagrams. However, enforcing these rules using such de facto standard as UML is still quite problematic. Principles of interplay among UML constructs, which define semantics of business processes and business data, are not completely clear.