This paper presents experiences and reflections from using the EKD Enterprise Modeling method since the beginning of the 1990'ies. A large number of application cases have been carried out. The paper focuses on the EKD modeling language, the EKD modeling process and supporting tools.
Enterprise Modeling (EM) has for many years been a central theme in information
systems engineering research. A number of different methods have been proposed.
There are two main reasons for using EM [1]: (
The remainder of this paper is organized as follows. In section 2 we present the EKD Enterprise Modeling method. Section 3 describes a number of applications of the method. Our reflections and experiences are presented in Section 4, while Section 5 discusses the findings and provides some directions for future work.
In Scandinavia, methods for Enterprise Modeling (EM) was initially developed in the 1980’s by Plandata, Sweden [21], and later refined by the Swedish Institute for System Development (SISU). A significant innovation was then the notion of business goals as part of an Enterprise Model, enriching traditional model component types such as entities and business processes. The SISU framework was further developed in the ESPRIT projects F3 – “From Fuzzy to Formal” and ELEKTRA – “Electrical Enterprise Knowledge for Transforming Applications”. The current framework is denoted EKD–“Enterprise Knowledge Development” [7, 12]. 1.1
EKD – Enterprise Knowledge Development method [7] is a representative of the Scandinavian strand of EM methods. It defines the modeling process as a set of guidelines for a participative way of working and the modeling product in terms of six sub-models, each focusing on a specific aspect of an organization (see table 1).
Goals Model Business (GM) Rules Model (BRM)
Concepts Model Business Process Actors and (CM) Model (BPM) Resources
Model (ARM) Focus Vision and Policies and Business Business
strategy rules ontology operations Issues What does the What are the What are the What are the organization business rules, things and business want to how do they “phenomena” processes? How achieve or to support addressed in other do they handle avoid and organization’s sub-models? information and why? goals? material? Com- Goal, prob- Business rule Concept, Process, po- lem, external attribute external proc., nents constraint, information set,
opportunity material set Table 1: Overview of the sub-models of the EKD method [22]
The modeling components of the sub-models are related between themselves within a sub-model (intra-model relationships), as well as with components of other sub-models (inter-model relationships). Figure 4 shows inter-model relationships. The ability to trace decisions, components and other aspects throughout the enterprise is dependent on the use and understanding of these relationships. For instance, statements in the GM need to be defined more clearly as different concepts in the CM. A link is then specified between the corresponding GM component and the concepts in the CM. In the same way, goals in the GM motivate particular processes in the BPM. The processes are needed to achieve the goals stated. A link therefore is defined between a goal and the process. Links between models make the model traceable. They show, for instance, why certain processes and information system requirements have been introduced.
While different sub-models address the problem domain from different perspectives, the inter-model links ensure that these perspectives are integrated and provide a complete view of the problem domain. They allow the modeling team to Technical Component & Requirements
Model(TCRM) Organizational Information structure system needs Who are What are the responsible for business goals and requirements to process? How are the IS? How are the actors they related to interrelated? other models? Actor, role, IS goal, organizational IS problem, unit, individual IS requirement,
IS component
assess the business value and impact of the design decisions. There are two alternative
approaches to notation in EKD: (
In order to achieve high quality results, the modeling process is equally important as the modeling language used. There are two aspects of the process: the approach to participation and the process to develop the model.
When it comes to gathering domain knowledge to be included in Enterprise Models, there are different approaches. Common approaches are interviews with domain experts, analysis of existing documentation, observation of existing work practices, and facilitated group modeling. EM practitioners and EKD method developers have advocated a participatory way of working using facilitated group modeling (see e.g. [7, 9, 23, and 24]). In facilitated group modeling, participation is consensus-driven in the sense that domain stakeholders “own” the models and govern their contents. In contrast, consultative participation means that analysts create models and domain stakeholders are then consulted in order to validate the models. In the participatory approach stakeholders meet in modeling sessions, led by a facilitator, to create models collaboratively. In the sessions, models are often documented on large plastic sheets using paper cards. The “plastic wall” (Figure 2) is viewed as the official “minutes”, for which every domain stakeholder in the session is responsible. [23] give two main arguments for using the participative approach, namely: 1. The quality of models is enhanced if they result from collaboration between stakeholders, rather than from consultants’ interpreting stakeholder interviews. 2. The approach involves stakeholders in the decision making process, which facilitates the achievement of acceptance and commitment. This is particularly important when modeling is focused on changing some aspect of the domain, such as e.g. its visions/strategies, business processes and information system support.
In a modeling session, the EKD process populates and refines the sub-model types
used in that particular session gradually and in parallel. When working with a model
type, driving questions are asked in order to keep this parallel modeling process
going. This process has three goals: (
Application case at Riga City Council (RCC), Latvia, during 2001-2002 within the FP5 IST programme project “Hypermedia and Pattern Based Knowledge Management of Smart Organizations” (no IST-2000-28401) See further [27, 28]. The objective of this project was to develop and deploy a knowledge management (KM) system. Hence, the purpose of EKD modeling was to develop a specification and an adoption plan for a KM system. The case was structured into a number of sub-projects taking place at various departments – the Drug Abuse Prevention Center, the Traffic Department, the School Board, the Municipal Police, the Department of Environment, and the Department of Real Estate. Each of these used EKD to elaborate and resolve specific issues related to KM. Across the cases ca 60 stakeholders, 4 modeling facilitators and 7 modeling technicians were involved. The results were later integrated in order to develop a KM strategy for the RCC. The Micrografx FlowCharter tool was used for documenting the models and the BSCW tool for communicating within the project.
Application case at Verbundplan GmbH, Austria, the consulting branch of the largest energy producer in Austria, took place during 2001-2002 within the FP5 IST programme project “Hypermedia and Pattern Based Knowledge Management of Smart Organizations”, see [27, 29]. Similarly to the RCC case, the purpose of EKD modeling was to establish the vision, KM process as well as to capture business requirements for a KM system. EKD modeling was performed in three sub-projects: repairing damages in hydro power plants, risk management, and project identification. The results contributed to establishing the corporate KM process and the KM system. Models were documented by the Micrografx FlowCharter tool and the modeling team communicated through the BSCW tool. At the later stages of the project models were part of corporate knowledge repository supported by the Requirements Engineering Through Hypertext (RETH) tool and its web-export functionality [30].
Application case at Skaraborgs Sjukhus (SKaS) during 2003-2006, within the project Efficient Knowledge Management and Learning in Knowledge Intensive Organizations (EKLär), supported by Vinnova, Sweden. SKaS is a cluster of hospitals in Western Sweden collaborates with primary care centers and municipal home care. The objective of the project was to develop a KM system and routines to support knowledge sharing among actors in the healthcare process (see further [16]). The purpose of EM was to develop a knowledge map that describes the contents in and structure of the knowledge repository. The knowledge map is in the form of an EKD Concepts Model. iGrafx Flowcharter was used to document the model. The specifics of this project was that although the resulting Concepts model could be considered as relatively small, it was refined numerous times and constantly updated throughout the project in order to reflect the stakeholders’ understanding of the knowledge domain. This model essentially serves as a “blueprint” for the knowledge repository at SKaS.
Apart from these projects, EKD and its earlier versions have been used in a number of smaller problem solving and organizational design cases at e.g. Strömma AB (S), Ericsson (S), RRC College (LV), Livani Sistrict (LV), and British Airospace (UK).
We have collected our experiences throughout more than 10 years of EKD application. Some are of a general nature and some are specifically related to the EKD modeling language, the EKD modeling process and tool support for the method. 4.1
In our experience EKD has the potential to provide good results in terms of high quality models, improved understanding of the problem among domain stakeholders, improved communication between stakeholders and personal commitment among stakeholders to the modeling result. However, it is fair to say that the achievement of these results is more resource-consuming than may appear at first, due to the perceived simplicity of the method. It may appear that anyone can use the method with a minimum of training and practical experience. Our experience as well as research (see e.g. [23]) has shown that this is a false perception. The following preconditions need to be fulfilled before using the method in a real life situation that has any real importance to the organization concerned: - The modeling team must be given a clearly stated mission to pursue. - Sufficient time and other resources must be allocated to the activity, for the project group and for other people in the organization to engage in the modeling work. - The modeling team must be given authority to design or re-design organizational as well as technical processes, procedures, concepts, and rules. - The team must be well-balanced in terms of knowledge about the problem at hand. - There is a skilled and experienced modeling facilitator available.
In addition to these conditions, each particular situation should be assessed in order to decide whether or not it is appropriate to use the method. We have found that the characteristics in Table 2 distinguish appropriate from inappropriate situations.
Appropriate situations: consensus-oriented organizational culture management by objectives when agreement among stakeholders needs to be ensured when reliable information is otherwise difficult to obtain (e.g. multiple stakeholder perspectives need to be consolidated, wicked or ill-defined problems) Inappropriate situations: authoritative organizational culture management by directives constant “fire-fighting” strong sense of hidden agendas trivial problem lack of skilful modellers Each model type in EKD has its particular focus. Depending on the application context, some become more heavily used than others. However, whichever the application the Goals Model, the Business Process Model, the Concepts Model and the Actors and Resources Model tend to dominate EKD usage. These sub-models answer the Why, How, What and Who questions that need to be asked about an enterprise regardless of situation, be it systems development, process development or strategic development.
Even though EKD has its own modeling language, it allows replacing the modeling language of one sub-model with a similar modeling language addressing the same modeling problem. It also allows adding sub-models. Such adaptations can be made as long as the inter-model relationships in EKD are kept intact. This feature is useful when the situation in general is appropriate for using EKD but it has specific needs with regard to modeling capacity that the method cannot cater for.
There are two alternative notations in EKD, one simple and one more semantically rich. In the main portion of our work we have used the simple notation in modeling sessions, due to the fact that the stakeholders involved have to a large extent not been experienced modelers. We have found this to be a successful approach. In fact, an experienced facilitator makes training of domain stakeholders unnecessary. The facilitator will instead introduce the ideas of modeling and the notation little by little. Some situations, however, require more formality. We suggest that such formality is introduced after the modeling sessions using interviews with the stakeholders.
We have not seen the feature of explicit inter-model links in other methods. EKD suggests that they are useful for ensuring the quality of models and for driving the modeling process forward and reasoning about the model. The facilitator can use them to validate the models and the decisions of the modeling group..
As for model quality, practitioners are mainly concerned with whether the set of
resulting models are coherent as a whole and that they are possible to implement.
[23]. This is our experience too. Also, depending on the project objectives an
empirical study [31] shows that a sub-set of the criteria of completeness, flexibility,
simplicity, understandability, integration, usability (implementability), correctness
apply to Enterprise Models in most cases. Sometimes sessions do not produce high
quality models. Nevertheless, they may still add value through the discussions among
participants. The EKD modeling process hence produces two kinds of useful results:
(
The most useful features of the EKD process throughout our work are: 1) the parallel development of sub-models using inter-model links and 2) the participatory approach to modeling. In our opinion modeling languages need to be combined with a suggested way of working. This is, however, seldom provided. EKD is an exception. The danger of a lacking process is that it may imply that modeling in practice is fairly simple. In our experience it takes a long time to become a skilled modeler, and also to become a skilled modeling facilitator. Training should not be taken lightly. An organization that plan to develop this competency should have a long-term strategy. More about competency requirements for modeling can be found in [23].
The planning an EM project/activity is also critical. It is highly desirable that method experts have a strong influence on selecting domain stakeholders for the modeling team. Once they have been chosen, they need to be prepared for what will happen during the sessions. This is particularly critical in organizations that are not used to modeling in general and particularly to modeling in a group. Before the modeling session each participant has to understand the objective of the modelling session, agree upon the importance of this objective, feel personally capable to contribute to a positive result, and be comfortable with the rest of the team (including the facilitator). More about preparing for EM can be found in [22]. 4.4
The EM process needs to be supported by tools. The tool requirements depend on the organization’s intentions (e.g. will the models be kept “alive”) and situational factors (e.g. the presence of skillful tool operators and resources). More on how to select and introduce EM tools in organizations is available in [32].
Group meeting facilitation tools, e.g. GroupSystems, are used to support modeling. They have become more sophisticated and popular. However, they still lack specific support for participative EM, e.g. for guiding the modeling process [33], or “close to reality” graphic resolution. We recommend using a large plastic sheet and colored notes to document the model during a modeling session (Fig.2). Then modeling can be set up in almost any room with a sufficiently large and flat wall. Also it, allows the participants to work on the model without disturbing each other. If a computerized tool and a large projection screen are used, the participants have to “queue” in order to enter their contributions. This usually slows down the creative process. In addition the “plastic wall” is also cheap and does not require technicians to set it up. atblsa atblsa
Risna documented in MS Visio
Figure 2: Modeling result – an EKD Goals Model after 10 hours of modeling with 12 domain experts and after documenting it in Microsoft Visio.
After the modeling session the models on plastic may be captured with a digital camera. If they are to be preserved, e.g. included in reports, posted on the intranet, it needs to be documented in a computerized modeling tool (see fig.2). This category of tools includes simple drawing tools and more advanced model development and management tools. In “stand-alone” projects only drawing support may be needed. If so, simple drawing tools such as Microsoft Visio and iGrafx FlowCharter have proven to be useful and cost-effective [1, 32]. In other cases, e.g. when enterprise models need to be communicated to large audiences or linked with existing information systems, more advanced tools should be used. In this category of tools we find e.g. Aris (IDS Scheer) and Metis (Troux Technologies). Apart from modeling tools EM projects need group communication and collaboration tools. We have successfully used Basic Support for Collaborative Work (BSCW) tool (Fraunhofer).
Business requirements for EM tools include integration of EM tools with MS Office, model visualization and presentation requirements (often in web-format) as well as reporting and querying requirements. We have also observed a growing need to connect models to information systems, thus making the models executable. An extended presentation of requirements for EM tools is available in [32]. 5
This paper has presented some experiences from using the EKD EM method for more than ten years. The experiences made will influence further development of the method and its supporting tools. One of the main weak points is that, at present, the method lacks a software tool that supports the deployment of the method. Relationships for Business Process Reengineering. In Proceedings of the 13th International Conference on the Entity-Relationship Approach, Manchester, UK. 14. Wangler, B., Persson, A., Johannesson, P. and Ekenberg, L. (2003), Bridging High-level Enterprise Models to Implemenation-Oriented Models, in Fujita, H. and Johannesson, P. (eds.), New Trends in Software Methodologies, Tools and Techniques, IOS Press 15. Niehaves, B. and J. Stirna. (2006) Participative Enterprise Modelling for Balanced
Scorecard Implementation, Proceedings of ECIS’06, Gothenburg, Sweden. 16. Stirna, J., Persson, A., and Aggestam, L. (2006), Building Knowledge Repositories with Enterprise Modelling and Patterns - from Theory to Practice, Proceedings of ECIS’06, Gothenburg, Sweden 17. Wangler, B. and Persson, A. (2002), Capturing Collective Intentionality in Software Development, in Fujita, H. and Johannesson, P. (eds.), New Trends in Software Methodologies, Tools and Techniques, IOS Press, Netherlands, pp 262-270. 18. Wangler, B., Persson, A. and Söderström, E. (2001), Enterprise Modeling for B2B integration, In International Conference on Advances in Infrastructure for Electronic Business, Science, and Education on the Internet, L'Aquila, Italy (CD-ROM). 19. Gustas, R., Bubenko, J. A. jr and Wangler, B. (1995). Goal Driven Enterprise Modelling: Bridging Pragmatic and Semantic Descriptions of Information Systems, 5th European Japanese Seminar on Information Modelling and Knowledge Bases, Sapphoro, May 1995. 20. Kardasis P., Loucopoulos P., Scott B., Filippidou D., Clarke R., Wangler B., Xini G. (1998). The use of Business Knowledge Modelling for Knowledge Discovery in the Banking Sector, IMACS-CSC'98, Athens, Greece, October, 1998. 21. Willars, H. (1988). Handbok i ABC-metoden. Plandata Strategi. 22. Stirna, J., Persson, A. and Sandkuhl, K. (2007). Participative Enterprise Modeling:
Experiences and Recommendations, Proceedings of CAiSE’07, Trondheim, Norway . 23. Persson A, (2001). Enterprise Modelling in Practice: Situational Factors and their Influence on Adopting a Participative Approach, PhD thesis, Stockholm University, ISSN 1101-8526. 24. Nilsson, A. G., Tolis, C. and Nellborn, C. (Eds.) (1999), Perspectives on Business
Modelling: Understanding and Changing Organisations, Springer-Verlag. 25. Persson, A. and Horn, L. (1996) Utvärdering av F3 som verktyg för framtagande av tjänsteoch produktkrav inom Telia,Telia AB, Sweden, Doc no 15/0363-FCPA 1091097. 26. Bergenheim, A., A. Persson, D. Brash, J. A. J. Bubenko, P. Burman, C. Nellborn and J.
Stirna (1998). CAROLUS-System Design Specification for Vattenfall AB . 28. Kaindl H., Hatzenbichler G., Kapenieks A., Persson A., Stirna J., Strutz G. (2001) User Needs for Knowledge Management, deliverable D1, IST project no. IST-2000-28401 HyperKnowledge, Siemens AG Österreich, Austria. 28. Mikelsons, J., Stirna, J., Kalnins, J. R., Kapenieks, A., Kazakovs, M., Vanaga, I., Sinka, A., Persson, A. and Kaindl, H. (2002). Trial Application in the Riga City Council, deliverable D6, project no. IST-2000-28401 HyperKnowledge, Riga, Latvia. 29. Dulle H., (2002) Trial Application in Verbundplan, deliverable D5, IST Programme project
IST-2000-28401 HyperKnowledge, Verbundplan, Austria. 30. Kaindl, H., Kramer, S., and Hailing, M., (2001) An interactive guide through a defined modeling process, in People and Computers XV, Joint Proc. of HCI’01 and IHM’01, Lille, France, pp 107-124. 31. Larsson, L., Segerberg R., (2004) An Approach for Quality Assurance in Enterprise Modelling, MSc thesis, Deptment of Computer and Systems Sciences, Stockholm University, no 04-22. 32. Stirna, J. (2001). The Influence of Intentional and Situational Factors on EM Tool
Acquisition in Organisations, Ph.D. Thesis, Royal Institute of Technology, Sweden. 33. Nurcan, S. and Rolland, C. (1999). Using EKD-CMM electronic guide book for managing change in organisations, in Proceedings of the 9th European-Japanese Conference on Information Modelling and Knowledge Bases, Iwate, Japan.