Research and Advances
Computing Applications

Integrated Modeling: the Key to Holistic Understanding of the Enterprise

Current business modeling methods are one-dimensional. Here, an integrated modeling approach helps decision makers study the enterprise as a whole from multiple perspectives.
  1. Introduction
  2. Challenges of Enterprise Modeling
  3. Proposed Solution: An Integrated Modeling Environment
  4. The Conceptual Architecture Behind MODELMOSAIC
  5. Addressing the Modeling Challenges in MODELMOSAIC
  6. Conclusion
  7. References
  8. Authors
  9. Figures

Decision makers who want to optimize costs and quality need a holistic understanding of the structure and operations of the enterprise. But what are the tools that will enable such holistic understanding? Information technologies, particularly integrated enterprise systems (ERP systems) that provide accurate, real-time, enterprisewide information, are one important approach. In this article, we hope to make a case for another method of attaining holistic understanding of the enterprise: integrated modeling and analysis. We introduce the area, describe major modeling challenges, present an integrated modeling environment to address those challenges, and identify future directions for researchers and practitioners based on our work.

Enterprise modeling methods and tools enable decision makers to represent, visualize, understand, communicate, redesign, and improve the operations of an enterprise with a focus on timeliness, cost, and quality. The synergistic combination of descriptive graphical models created using enterprise modeling methods such as IDEF with the use of analytic methods such as optimization can deliver substantial results. A case in point is American Express, which used process modeling and analysis tools to offer new financial products and services with significant improvements in time to market. However, despite some success stories, technological advances in enterprise modeling and analysis have yet to filter into the mainstream of managerial decision making [11]. While information systems are beginning to realize the promise of integration (given the increasing use of ERP systems), modeling practices are not. Modeling methods and tools, though potentially valuable, are seen as complex, nonintegrated, time-consuming, expensive, and usable only by experts who are difficult to find [4, 8].

Enterprise modeling [10] is concerned with the representation and specification of various aspects of the enterprise operations—functional (what is being done, and in what order), informational (which and how many agents are used, required, or processed), resource (what or who carries out tasks and what policy applies), and organizational (the framework within which things are done). Enterprise analysis is concerned with turning those descriptive enterprise models into executable forms (such as discrete event simulation, optimization, queueing, and activity-based costing) through which different aspects of the enterprise can be studied dynamically.

Much has been accomplished in enterprise modeling and analysis over the last two decades. Researchers worldwide have proposed a wide range of modeling methods: CIMOSA, PERA, ARIS, CimTool, FirstSTEP, IBM’s FlowMark, to name a few. These methods were designed to represent functional, informational, resource, and organizational aspects of the enterprise. They advanced the much-needed representational power of the enterprise models by using set-theoretical and graphical constructs. Each new modeling method evolved to cover aspects inadequately addressed by previous methods. For a discussion of these methods, see [2, 3, 7]. In response to the increasing number of methods, in 1998 an IFIP-IFAC task force proposed the Generalized Enterprise Reference Architecture and Methodology (GERAM) [6] as an overarching framework for describing the components needed in all types of enterprise engineering and enterprise integration efforts. Enterprise modeling is a critical component of GERAM.

However, as a high-level abstraction, GERAM does not go down to the implementation level. Despite progress and attempts at a holistic approach, enterprise modeling efforts remain fragmented at the implementation level, as most models developed today do not talk to one another. The ones that do, do so in an ad hoc manner via an export-import interface. Our view of integrated modeling goes beyond merely having models talk to one other; it includes the consistency, integrity, and validity of all the models representing the enterprise. This requires an integration infrastructure that automatically propagates the changes made to one model type to all other model types. ERP systems achieve similar integration for transactions at the operational level. We need an ERP-style integration infrastructure for models. Only then can we achieve true integration of enterprise models, which will enable a holistic, consistent, and coherent picture of an enterprise from multiple perspectives. This vision is illustrated conceptually in Figure 1, using three exemplary perspectives of the enterprise.

Back to Top

Challenges of Enterprise Modeling

Heterogeneous methods and tools. Different conceptual enterprise models focus on different aspects of an organization, such as data, activity, process, and organizational structure. For example, a process modeling method focuses on the processes or tasks that are performed, on the partial ordering of these tasks with respect to time, and on the objects involved in each task. A data model, on the other hand, captures information about the types of data stored and managed in the enterprise, the characteristics of the data, and the relationships and constraints between them. Each type of model is necessary to describe a particular aspect of an enterprise; but to understand the entire enterprise it is necessary to go beyond individual models. Understanding the whole requires understanding not only the parts, but the relationships between them. Depending on how we conceptually slice the enterprise, the parts can be seen at several levels and may be organized into different categories. Hence, one practical challenge is how to integrate different types of models, given the wide variety of methods used to create them.

Model correlation. Decision makers need to correlate descriptions and analyses from different models to develop a reasonably complete and coherent picture of the enterprise. Such correlations will enable them to detect conflicts and inconsistencies between models, identify missing information, and calculate the effect of changes in one aspect of the enterprise on other aspects. Because each type of model is generally captured, represented, and stored using a standalone application, the challenge is to explicitly correlate enterprise models and to evolve them over time while maintaining consistency between them.

Representation extensibility. While modeling an enterprise, decision makers and domain experts are limited by the representational capabilities of the particular modeling method and tool. Information specific to a particular domain or falling outside the scope of the method can only be captured and stored using ad hoc notes and annotations (if the modeling tool supports such annotations). For example, if you wanted to add a new capability to a function modeling method in order to capture the priority of the activities within a process, you would have to record the priority in annotations of that activity object. Though beneficial, this representation is not extensible, nor can it be available to other modeling methods and tools. The challenge here is to make enterprise models truly extensible.

MODELMOSAIC provides the necessary functionality for capturing and manipulating the information relevant to particular model types by enforcing the principles of each model type in the form of business rules.

Enterprise model compiling.Traditionally, executable models of enterprises (such as simulation or optimization models) are built from scratch using a modeling tool to analyze a problem or to take advantage of an opportunity. Once the analysis is completed, the model is discarded with no thought to additional use, because it was specific to the problem being analyzed, and the chance of the same problem repeating under the same business conditions is very small. This single-use, throwaway mentality toward enterprise executable models is expensive, time-consuming, and wasteful [4,1]. Therefore the challenge is to automate the creation of executable models from the descriptive representation of the enterprise on an as-needed basis.

Back to Top

Proposed Solution: An Integrated Modeling Environment

These challenges can be addressed by an integrated modeling environment that supports the capture, representation, storage, and interlinking of a variety of model types. Specifically, the environment should support the development of all types of models needed to capture the various aspects of an enterprise; the seamless integration and correlation of these models; extensible model types that can capture and represent additional information content; and the automatic creation of executable models of the enterprise from a comprehensive set of descriptive representations. Despite the tremendous interest in enterprise modeling and analysis in recent years, no single modeling methodology or environment possesses these capabilities. To address this need, we have developed a conceptual framework for an integrated modeling environment and used it as the basis for a software application called MODELMOSAIC. This application provides an integrated suite of enterprise modeling tools and an extensible environment for enterprise analysis. Figure 2 is a screen shot of the MODELMOSAIC toolkit. It shows a number of different viewpoints of the enterprise in their abstract forms using function, process, and simulation models.

MODELMOSAIC provides the necessary functionality for capturing and manipulating the information relevant to particular model types by enforcing the principles of each model type in the form of business rules. The information captured in various model types, which constitute the enterprise model set, is stored in a single integrated information base. These models can be viewed and manipulated concurrently via a common graphical user interface. The environment provides a mechanism for reusing information captured in one model type to create other model types. This is done by identifying the relationships among different information objects supported by various model types. These relationships are represented in the form of business rules, and are used to automatically generate new perspectives (descriptive or analytic models) of the enterprise from one or more of the existing models of different types.

Back to Top

The Conceptual Architecture Behind MODELMOSAIC

Figure 3 illustrates the conceptual system architecture that drives MODELMOSAIC. The heart of the system, the core component, provides a user interface shell, database access services, and mechanisms for model integration. The user interface shell is the foundation for the common user interface, which includes common menus, toolbars, and other GUI objects. The database access services store and manage models (and other information types) in the integrated database structure. Finally, a powerful inference engine provides model integration and inter-model consistency through automatic change propagation.

The first echelon of the architecture contains the various modules, called modelers. Each modeler supports the creation and management of a single model type. The modelers are separate, complete components that interface with the core through the common database and common user interface. They provide it with a description of the types of information objects that must be managed in the common database. Each modeler requests the core component to create the necessary database tables and entries to manage those information objects. This modular construction allows new modelers to be added to the environment without disturbing the structure of the database.

At the second echelon, the plug-in applications provide another level of functionality and integration. They extend the environment by building on the core component and one or more of the modelers. This echelon facilitates the development of enterprise model compilers to create bridges between the basic modelers and a variety of analysis tools.

At the third and final echelon lie the custom extensions. Custom extensions are similar to the applications and are melded with the environment in a similar way. The goal is to make it easy for end users to extend the modeling environment. This echelon gives the environment virtually unlimited extensibility while maintaining the integrity and consistency of the formal methodologies supported by the tool.

Back to Top

Addressing the Modeling Challenges in MODELMOSAIC

MODELMOSAIC addresses model correlation challenges because it can capture, represent, and store intermodel relationships and use these relationships to maintain consistency across the enterprise model set. Users can identify and record relationships between model elements across model types. Because the integrated modeling environment enables users to view multiple models concurrently, they can identify relationships across models by selecting the model elements to be related and the relationship between them. Once identified, the relation instance is stored in the integrated information base as a business rule and is used to maintain consistency between the related elements.

There are several modeling tools in the marketplace today, but most of these either use proprietary methods or allow the use of extant modeling methods but in a standalone manner with minimal integration across methods.

Representation extensibility. MODELMOSAIC enables extensions (new information types) to be added and seamlessly connected to the existing model types through the underlying integrated information base. The main structure of the integrated information base is a flexible knowledge representation schema called the Container Object System (COS) [9]. The COS representation scheme is based on the separation of existential knowledge from descriptive knowledge. A container object has a non-qualitative, unique property that distinguishes it from any other objects within the environment. The existence and identity of the object are determined by that property. A set of descriptive information elements for an object, including the object’s qualitative properties and its relationships to other objects, forms a “container” that is attached to the object itself and can easily be modified over time.

The integrated information base represents and manages various information types needed by the various modeling methods in the form of containers. An object can have multiple containers that describe it in different perspectives. This approach enables both extensions to the properties of existing information types (adding a property or relation to a type of container) and the creation of new information types (construction of new types of containers).

Enterprise model compiling. The extensibility of the information base allows information specific to analysis methods to be captured, represented, and stored in the information base as well as explicitly linked to the appropriate model elements. In this fashion, information captured for the purpose of generating a particular executable model can be reused for other analysis efforts. The integrated information base also enables enterprise model compilers to generate executable models from the total set of enterprise models. In this integrated modeling environment, because the various models that constitute the enterprise model set are tightly coupled through meaningful relationships and constraints, executable models can be generated from a comprehensive picture of the enterprise.

Back to Top


We have analyzed the key challenges facing a broader use of enterprise modeling and analysis methods, most of which stem from a lack of integration among heterogeneous modeling and analysis methods and tools. There are several modeling tools in the marketplace today, but most of these either use proprietary methods or allow the use of extant modeling methods but in a standalone manner with minimal integration across methods. These modeling tools do not provide automated execution model compiling capabilities, making them merely capture tools for system description. Hence, they do not adequately address the challenges of model correlation, representation extensibility, and model compiling in a single environment. These challenges have contributed to the perception that enterprise modeling and analysis is complex, time-consuming, expensive, nonintegrated, and usable only by specialists. There have been several efforts at integration, some aimed at high-level standards such as GERAM, some toward broad-based enterprise architectures such as CIMOSA, some toward developing common modeling languages such as UML, and some within individual communities of interest such as the XML-based Business Process Management Initiative (

The next generation of ERP II systems should integrate enterprise modeling and analysis seamlessly so that enterprise models can exchange information with the ERP system as needed.

Our research complements existing commercial and research approaches and represents a promising attempt to address the integration challenge at a practical level. We have described the conceptual approach and the implementation of MODELMOSAIC. Several future research directions are important. First, further work is needed to refine the theoretical basis for integrating extant enterprise modeling and analysis approaches. Second, there is a need for a completely Web-based integrated modeling environment for distributed collaborative users, which should help address some of the modeling challenges we have presented in this article. Third, modeling integration needs to extend beyond the enterprise to cover multiple cooperating enterprises in a supply chain. This should parallel developments in newer ERP II systems [12] that support the extended enterprise with modules for customer relationship management, advanced planning systems, supply chain management, and collaborative commerce.

Finally, integration for enterprise modeling and enterprise information should not be seen as two separate problems. The next generation of ERP II systems (which we call ERP III systems) should integrate enterprise modeling and analysis seamlessly so that enterprise models can exchange information with the ERP system as needed. This will be a quantum jump from the graphical process models in current ERP systems, which at their best are merely user interfaces to underlying process functionality. In the long run, researchers and practitioners working together at the levels of standards, methods, theory, and tools can realize the dream of true integration to achieve holistic understanding of the enterprise.

Back to Top

Back to Top

Back to Top


F1 Figure 1. An example of an integrated enterprise model set.

F2 Figure 2. A screen shot from the MODELMOSAIC software environment.

F3 Figure 3. Multi-layered system architecture.

Back to top

    1. Benjamin, P., Erraguntla, M., O'Brien, T., Delen, D., and Ramachandran, S. An intelligent assistant for optimization modeling. In Proceedings of the Eighth Industrial Engineering Research Conference. (Phoenix, AZ, 1999), IEEE, 289–298.

    2. Berio, G., and Vernadat, F.B. New developments in enterprise modeling using CIMOSA. Comput. Ind. 40 (1999), 99–114.

    3. Curtis, B., Kellner, M. I., and Over, P. Process modeling. Commun. ACM 35, 9 (Sept. 1992), 75–90.

    4. Delen, D., Pratt, D. and Kamath, M. Enabling multi-use, multi-tool models of manufacturing systems through an experimental frame expert system. J. Intell. Manuf. 12, 3 (2001), 247–255.

    5. Delen, D. Development of an Expert System Based Experimental Frame for Modeling of Manufacturing Systems. Ph.D. Dissertation, School of Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, 1997.

    6. GERAM (1999). Generalized Enterprise Reference Architecture and Methodology, Version 1.6.3. IFAC-IFIP Task Force on Enterprise Integration;

    7. Kamath, M., Dalal, N. P., Chaugule, A., Sivaraman, E. and Kolarik, W. J. A review of enterprise process modeling techniques. Scalable Enterprise Systems: An Introduction to Recent Advances. V. Prabhu, S. Kumara, and M. Kamath, Eds..Kluwer Academic Publishers, Boston, MA, 2003, 1–32.

    8. Mize, J. H., Bhuskute, H.C., Pratt, D.B. and Kamath, M. Modeling of integrated manufacturing systems using an object-oriented approach. Inst. Indus. Eng. Trans. 24, 3 (1992), 14–26.

    9. Sanders, L. K., Browne, D., Menzel, C., and Mayer, R. J. Container objects: a description based knowledge representation scheme. In Proceedings of Autfact `91, (1991), 155–163.

    10. Vernadat, F. B. Enterprise Modeling and Integration. Chapman & Hall, London, 1996.

    11. Whitman, L. and Huff, B. On the use of enterprise models. Int. J. Flexible Manuf. Syst. 13 (2001), 195–208.

    12. Zrimsek, B. (2002). ERP II: The Boxed Set;

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More