Central to Multi-View Modeling Methods (MVMMs) is the structuration of an enterprise and its underlying systems into views and levels in order to curtail their complexities. However, efficient application of MVMMs considerably depends on the availability of corresponding modeling tools- the so-called Multi-View Modeling (MVM) tools. The latter are crucial for machine processing of models (views) and hence for managing inter-view consistency. This is the reason why the focus of this paper is geared towards developing a MVM tool for the Enterprise Knowledge Development (EKD) method - a representative of MVMMs. This tool is intended not only to fill a research gap, but also to implement insights and findings derived from our previous work on inter-view consistency. Two different ways of managing consistency are indeed stressed: consistency by construction and consistency checking. Compared to the latter, the former advocates preventing inconsistencies rather than handling/correcting them. The tool has been implemented using ADOxx platform and realized within the Open Models Laboratory (OMiLAB).
At the core of Multi-View Modeling Methods (MVMMs) is Multi-View Modeling
(MVM). MVM is a widely accepted technique to reduce the complexity of a system
under study (e.g., a business process, enterprise, cyber-physical system). The key idea
behind this technique is to capture different aspects of the modelled system (e.g. its
structure and its behavior) by different views (models) [
Nonetheless, efficient application of MVMMs considerably depends on the
availability of corresponding modeling tools [
In its current version, the tool focuses on the process related aspects of EKD since
this method advocates a MVM approach for business processes. An EKD business
process model is thereby depicted in terms of three complementary views (sub-models),
where each view captures one or more perspectives (aspects). Based on this, these
views are not independent from each other. There is hence a need for managing
consistency among them. The proposed tool is intended not only to fill a research gap, but
also to implement insights and findings derived from our previous work on inter-view
consistency [
This paper is organized as follows: Section 2 provides the theoretical foundations for this work and briefly presents related works. Section 3 puts emphasis on the conceptualization of the EKD modeling method. In Section 4, an overview of the EKD modeling tool along with a case study demonstrating the application of this tool are given. Lastly, section 5 concludes the paper with an outlook on future research. 2 2.1
Before presenting the motivation for this work, key terms like viewpoint, view, concern
and consistency need to be defined. By viewpoint, we refer to the modeling language
used to specify a view which is represented by a conceptual model. Each view allows
capturing perspective(s). The latter is used to denote certain aspect(s) from which the
system under study can be viewed (e.g., its behavioral and structural aspects). A
perspective can then be a synonym to the term 'concern' defined by ISO/IEC/IEEE 42010
1 http://omilab.org
as "any interest in the system" [
Inter-view consistency is an important aspect on which depends the quality of the
created views and, hence their capability to be involved in the “design by reuse”.
Inconsistent views are indeed not only hard for their users to grasp but also hamper the tasks
of process analysis, redesign, reuse, and automation [
Following the findings of a systematic literature review of inter-view consistency
[
Two reasons constitute the motivation behind this work. First, the support of the EKD method by a tool to foster its efficient application. Second the implementation of the two diverging ways of managing consistency. Such implementation is needed to justify the preference of one way over the other. 2.2
The development of EKD method started in late nineties within a research project called ELECTRA (ELectrical Enterprise Knowledge for TRansforming Applications), in which several researchers have been involved including the co-author of this paper. EKD stands for a framework for modeling enterprise knowledge that, grounding on the terms "dimension” and "perspective", it allows mastering not only the complexity of the enterprise as a whole, but also that of business processes. As exhibited in Fig. 1, the EKD method structures the enterprise in three key dimensions: Goals, Business processes and Information systems. In its current version, the tool targets the dimension "business processes" which is organized into three perspectives (see Fig. 1): Actors/ Roles, Roles/Activities and Business Objects. Each perspective is captured by a view (i.e., sub-model). Consequently, according to the EKD framework, a business process model is divided into three views: actors/roles view, roles/activities view and business objects view. A horizontal separation of concerns is therefore applied by this method.
To facilitate the implementation of the EKD modeling method in a modeling tool,
we perceived this method as a specialized instantiation of the generic modelling method
framework as introduced in [
The Semantic Object Model (SOM) tool [
Compared to existing MVM tools, the aim pursued in this paper is to develop a MVM tool for the EKD method that implements these two diverging ways of managing consistency. Such implementation is needed for evaluating them with respect to quality attributes like time and consistency, and therefore for providing a reasonable reason to prefer one over the other. 3
This section is devoted to defining the language for EKD business process models. The language is specified by a metamodel. As shown in Fig. 2, the EKD business process metamodel highlights concepts and their relationships that are used for modeling business processes. It is specified using UML class diagrams. Concepts are linked by an association relationship along with cardinalities to denote how many instances of the one concept can be connected to one instance of the other concept. The metamodel encompasses three interrelated viewpoints: actor/role, role/activity, and business objects. An EKD business process model is then represented in the form of three complementary views (sub-models). Each view is specified by its corresponding viewpoint (a metamodel part).
Fig. 2. The EKD business process metamodel [
The actor/role viewpoint is about representing the organizational perspective of a business process. The latter focuses on where and by whom activities are performed. The concepts "Actor" and "Role" are consequently at the core of the actor/role viewpoint. Here, it is worth noting the distinction between these two concepts. An “Actor” refers to a physical enterprise entity (e.g., customer, business unit manager, financial director), whereas the “Role” defines the responsibility assigned to an actor within a business process (e.g., loan request validator). An actor plays one or more roles. Each "Role" is responsible of one or more operational goals. Moreover, there is a "Dependency" between roles. Such dependency can be of type "Authorization", "Coordination", "Objective" or "Resource". Fig. 3 exhibits the representation of an actor/role view.
Fig. 3. Representation of an actor/ role view.
By contrast to the actor/role viewpoint, the role/activity viewpoint puts emphasis on the functional and behavioral perspectives of a business process. Functional perspective represents what activities should be performed. While, behavioral perspective represents when and how activities are performed. Central to role/ activity viewpoint are "Activity" and the control flow: "Sequence", "Parallel" and "Alternative". A representation of parallel and alternative executions is given in Fig. 4.
Last but not least, the business objects viewpoint that is articulated around the concept "Object". Each business activity of a given business process requires business objects as generations of input and output data. A business activity therefore use/produce business objects. Business objects refer to physical and informational elements of an enterprise that are required for its functioning. This viewpoint highlights then the informational perspective depicting the business objects handled by a process, their structures, and the links between them. As outlined in the metamodel (Fig. 2), an "Object" has one or more "State" and is composed of one or more "Operation". An "Operation" causes one or more "Transition". Accordingly, a business objects viewpoint allows describing not only a static aspect of business objects (i.e. business objects and their links), but also a dynamic one enabling representing business objects lifecycles according to activities carried out within business processes. Note that the concept "Business process" appearing in the metamodel denotes a compound building block within a business process model. It is therefore not a basic concept of the language.
This section discusses the conceptualization of the theoretical foundation introduced above towards a modeling tool for the EKD method. The tool is based on the ADOxx metamodeling platform. In this section, an overview of the tool functionalities as well as a case study demonstrating the use of the tool are given. 4.1
The EKD tool enables modeling of the EKD business process models (the second
dimension of the EKD enterprise framework on Fig. 1). Moreover, the tool allows to
carry out the two principles of MVM (cf. section 2.1). Views can therefore be created
following either the system-oriented or the diagram-oriented MVM. The modeler is the
one who decide which principle to follow. Table 1 brings an insight on how the EKD
viewpoints have been implemented in ADOxx.
The implementation of this functionality requires as prerequisite the definition of the
corresponding consistency rules. The latter have been defined in a previous work [
In the system-oriented MVM, the system (i.e. the tool) plays a key role in managing inter-view consistency. It somehow constructs consistency and hence relieves the modeler from the burden of handling/correcting inconsistencies. By contrast, in the diagram-oriented MVM, it is the modeler who plays a key role in managing inter-view consistency as it is up to him to resolve inconsistencies once detected by the system. The consistency by construction functionality is supported by the event handlers in ADOxx along with the attribute type "interRef" which together promote an automatic consistency preservation. Event Handlers are AdoScripts that are executed when certain events occur (e.g. create a node, create a connector, edit an attribute value,etc). "InterRef" is an attribute type that stands for a reference on a model or an instance. One example of EKD consistency rules that have been implemented in this way is rule 1 (roles in the actor role view and those in the role activity view have to be the same). The corresponding AdoScript implementing this rule is shown in Fig. 5. Fig. 5. Consistency by construction: implementation of rule 1 using an event handler (after_create_modeling_node).
The implementation of other rules has taken the form of guidelines for the modeler. Example of these rules is rule 2 (an operational goal has not to be atomic). An excerpt AdoScript implementing this rule is presented in Fig. 6. The aim always is to prevent inconsistencies as much as possible. As to the consistency checking functionality, it refers to an AdoScript code that allows comparing the created views based on the defined consistency rules. As a result of executing such functionality, a list of consistency issues is displayed to the modeler who has to handle them. For some EKD consistency rules, the execution of this functionality requires a system/modeler interaction using “EDITFIELD” in AdoScript as is the case for rule 2 (see Fig. 7). This subsection demonstrates the use of the EKD tool for modeling a cab booking process. In this example, we give a brief idea of the execution of the two diverging functionalities of managing inter-view consistency: consistency by construction and consistency checking. The execution of the AdoScript implementing rule 2 according to the consistency by construction approach (see Fig. 6) for the cab booking process gives rise to a guideline (see Fig. 8). Fig. 8 shows that the applied consistency rule (i) is of type semantic, (ii) concerns the role “Cab booking requester” and (iii) is implemented as a guideline. The latter insists on the necessity of translating the operational goal “Book a taxicab” into at least two activities in the corresponding role activity view.
The execution of the consistency checking functionality gives rise to the output presented in Fig. 9. As shown in this figure, such execution allows the detection of a set of consistency issues found in the created views. Each consistency issue is accompanied with its nature (syntactic or semantic) and the modeling element that is source of this issue. The paper at hand reports on a MVM tool for the EKD enterprise modeling method. In its current version, the tool accounts for two different ways of managing inter-view consistency: consistency by construction and consistency checking. In contrast to the latter, the former advocates preventing inconsistencies rather than correcting them. The implementation of these two ways in one tool is needed for evaluating them with respect to quality attributes like time and consistency, and therefore for providing a reasonable reason to prefer one over the other. The aforementioned ways of managing inter-view consistency dovetail respectively with two different principles of MVM: system-oriented MVM and diagram-oriented MVM. The tool has been implemented using ADOxx platform and realized within the OMiLAB. The development of this tool is an ongoing work. Therefore, we intend to finish the development of the tool and evaluate it with students to determine its usefulness and to improve it accordingly.