Article Preview
TopIntroduction
Information systems can be conceptualized by using graphical constructs of semantic modeling languages. Such conceptualizations are intrinsically complex engineering products. They are typically represented and visualized across disparate modeling dimensions (Zachman, 1996). Conceptual representations are built fragment by fragment on different levels of abstraction. Unfortunately, system designers still do not have effective methods for detecting semantic inconsistency, incompleteness and ambiguity of specification fragments. One of the difficulties resides in the information system analysis and design tradition to conceptualize separately organizational (business and service) architecture, data architecture, application architecture and deployment architecture. Semantic integrity of different diagram types is a fundamental problem in the traditional information system analysis and design methodologies. Verification of integrity between business process and business data is especially difficult. It is recognized that the Unified Modeling Language (UML) support for such task is quite vague, because integration principles of different diagram types are still lacking (Harel and Rumpe, 2004).
Conceptualizations cannot be interpreted by stakeholders in two different ways. Disambiguation of design should be regarded as driving force in conceptual modeling. More abstract conceptual representations must be suitable for inspection by business managers who have no technical background. Conceptual modeling methods should also help to verify semantic inconsistency and incompleteness of representations on various levels of abstraction. Current conceptual modeling practices are not quite ready for the achievement of such goals. Typically, conceptual representations help system developers (Kung & Solvberg, 1986):
- 1.
To increase understanding of domain,
- 2.
To serve as a common basis for design,
- 3.
To facilitate communication between system analysts and designers,
- 4.
To serve as specification of system requirements.
Even if the listed advantages are taken into account, it is recognized that good quality of conceptual modeling (Burton-Jones & Meso, 2006) is very difficult to attain. This situation makes information system analysis and design more art than science. The semantic integrity problems of conceptual representations create difficulties in communicating the semantic details of application domain among stakeholders. Conceptual specifications are often ambiguous, incomplete and inconsistent. They fail to serve as a basis for reaching consensus among system designers. The consequence is a communication gap between business analysis experts and system designers.
There are many concrete recommendations given by theoreticians for improving the existing conceptual modeling methods (Wand et al., 2000) and practices (Evermann & Wand, 2005). Typically, such improvements are either misunderstood or ignored by practitioners, because they are dealing with the partial solutions of semantic integrity problems. The consequence is that conceptual representations are still very difficult to use in information system analysis and design. There are many reasons why conceptual modeling fundamentals fail to provide a strong foundation of information system development. Conceptual modeling is still an emerging discipline. In this article we analyze several theoretical enhancements to achieve semantic integrity of conceptualizations. The proposed solutions are illustrated by using our enterprise modeling (EM) approach (Gustas & Gustiene, 2008). It helps us to theoretically investigate the semantic power of various diagram types. This article makes an attempt to discuss the following difficult questions: