An Architecture of Ontology-aware Metamodelling

Platforms for Advanced Enterprise Repositories

Srdjan Zivkovic, Harald Kühn and Marion Murzek

BOC Information Systems GmbH, Wipplingerstrasse 1, 1010 Vienna, Austria

Abstract. Enterprise repositories have become core information assets of to-

day’s enterprises. They store and manage models of different aspects of an en-

terprise, such as business strategy, business processes, organizational struc-

tures, and IT infrastructure in an integrative way based on a bundle of domain-

specific modelling languages. To produce such models for enterprise repository

and profit from their usage, number of mechanisms such as model querying,

validation, simulation, model transformation, versioning and traceability are

needed. Metamodelling platforms provide flexible and extensible environment

for realizing such advanced enterprise repositories, where various mechanisms

working on repository may be integrated, thus providing a superior modelling

solution. Going one step further, the question is, can such platforms profit from

semantic technologies? How architecture of ontology-aware metamodelling

platforms can be designed? In this paper, we give an in-depth view of such plat-

form architecture by discussing its main building blocks and their dependen-

cies.

1 Introduction

Comprehensive enterprise modelling tools are for enterprise repositories as database

management systems are for raw data. Raising the abstraction level of data manage-

ment, modelling repository-enabled tools offer sophisticated means for storage, que-

rying, visualising and manipulating enterprise information, thus leading to a model

management. Such tools enable a holistic approach to IT-based, model-aware strat-

egy, business process and IT management of the enterprise. Metamodelling platforms

provide flexible environment for realizing enterprise repository based modelling

tools. Such platforms offer: (1) modelling and metamodelling capabilities (2) set of

mechanisms to work on models and metamodels (3) guidance on how to apply par-

ticular modelling method using corresponding mechanisms in order to achieve

method specific goals [2]. Here, a model as a data abstraction represents a first-order

entity for capturing enterprise information. Going one data abstraction step further,

the question is, can enterprise modelling tools become ontology-aware? How can

ontology and related semantic technology leverage models & metamodels, mecha-

nisms and guidance of the modelling process? We analysed some scenarios on how

ontologies and automatic reasoning may bring benefit to metamodelling platforms

[2]. We concluded that modelling languages and models enriched by machine-

Zivkovic S., Kühn H. and Murzek M. (2009).

An Architecture of Ontology-aware Metamodelling Platforms for Advanced Enterprise Repositories.

In Proceedings of the Joint Workshop on Advanced Technologies and Techniques for Enterprise Information Systems, pages 95-104

DOI: 10.5220/0002202900950104

 SciTePress

readable semantics enhance quality of modelling solutions. Furthermore, configura-

tion of mechanisms may be semi-automated and their quality of service may be en-

hanced by relying on ontologies. Finally, the guidance gains benefits from ontologies

by having process models extended by formalized rules, conditions and actions, thus

leading to more flexible guidance. Considering previous findings, we extended the

metamodelling architecture to be ontology-aware.

Consequently, in this paper, our main goal is to give an in-depth view of such

platform architecture by discussing its main building blocks and their dependencies

from the logical viewpoint. We also summarize our current results of the work-in-

progress and give outlook for the future work.

2 Logical Architecture

The ontology-aware platform architecture for advanced model repositories extends

the generic metamodelling platform architecture [1]. It introduces the ontology as-

pect, such that the platform core features may profit from semantic technologies [2].

The root core architectural element is the meta-metamodel which contains the

concepts available for the definition of modelling languages. Based on it, the meta-

model library contains metamodels of defined modelling languages. The metamodel

library conforms to the meta-metamodel and, in turn, forms the foundation of the

model repository, where all models are stored. As an extension to models on different

levels, the ontology repository serves as storage of the domain, upper-level and proc-

ess ontologies. These four elements form the ontology-aware model management

building block (see section 2.1). All mechanisms used for evaluating and processing

of models, metamodels and ontologies are part of the extensible ontology-aware,

model-aware mechanisms building block (see section 2.2). Persistency services ab-

stract from concrete data persistence systems and offer support for the durable storage

of models, metamodels, and ontologies. Access services serve two main tasks (section

2.3). On the one hand they enable the open, bi-directional exchange of all metamodel,

model and ontology information. On the other hand they cover all aspects concerning

security such as access rights, authorization, and en-/decryption. User interface com-

ponents such as graphical and textual based editors are used for the definition, repre-

sentation and maintenance of models, metamodels, ontologies and mechanisms (see

section 2.4). A workbench layer serves as a common environment for integrating

different editors in a single modelling solution.

Persistency Services

Model Repository

Model, Metamodel, Ontology and Mechanism Editors

(Textual, Graphical etc.)

Metamodel Library

Model-aware and

Ontology-aware

Mechanisms

Access Services

Meta-metamodel

Ontology-aware MDE Workbench

Ontology

Repository

Ontology-aware Model Management

Fig. 1. Platform Architecture for Advanced Enterprise Model Repositories.

2.1 Ontology-aware Model Management Building Block

Meta-metamodel. Given the 4-layer metamodelling architecture, the meta-

metamodel defines the constructs available for the definition of modelling languages.

Typical metamodelling constructs are class, relation, endpoint, attribute, model type,

package etc. There are different possible implementations of the M3 model for differ-

ent modelling frameworks. The Meta Object Facility (MOF) is a standard meta-

metamodel to build the OMG-based family of modelling languages like UML and

BPMN [3]. Another implementation of the M3 model is the Ecore model [4], the

meta-metamodel of the Eclipse Modelling Framework (EMF) which is, from its cov-

erage, comparable to the subset of MOF, essential MOF (EMOF). Ecore with its

features is customized to support seamless projection to Java. A further implementa-

tion of the M3 model is the KM3 [5], the meta-metamodel of the ATLAS Model

Management Platform [6]. The KM3 is the lightweight textual metamodel definition

language which allows easy creation and modification of metamodels. It uses just few

concepts like Class, Attribute and Reference. It is structurally close to EMOF and

Ecore. Comparably, the ADOxx Metamodelling Platform

implements on M3 level

the ADOxx Meta

-Model, which is optimized for the rapid definition of the visual

modelling languages for enterprise modelling.

Metamodel Library. A metamodel library represents a collection of metamodels of

defined modelling languages. A metamodel describes basic constructs of a modelling

language. A metamodel is defined using meta-metamodel constructs. Metamodel

forms the foundation for the model repository, where all models conforming to the

corresponding set of metamodels are stored. For example, the BPMN metamodel

alone or a family of domain specific languages (DSL-s) which cover different aspects

of an enterprise are all defined on the M2 layer. Using a translational semantics

ADOxx

is the extensible, multi-lingual, multi-os, repository based platform for the development of

modelling tools of the BOC Group. ADOxx

is a registered trademark of BOC Group. http://www.boc-

group.com.

M3 Layer

M2 Layer

(Metamodel

Library Layer)

M1 Layer

(Repository

Layer)

Fig. 2. Ontology-aware Model Management Building Block.

approach, the semantics of some language L can be translated to some other language

L1 with known semantics. Thus, models can be transformed to other languages

whose semantics are formally defined, such as ontologies. Therefore, an ontology

language like ODM [7] is defined on M2 level as well. Ontology languages and target

modelling languages are then combined on the M2 level, which all results in an inte-

grated modelling approach on M1 level, where semantic constraints on models may

be expressed using ontology concepts (see Fig. 2). The TWOUSE approach addresses

the integration of UML-like languages and OWL, which results in an integrated mod-

elling language for software engineers [8]. The same approach may be applied to an

enterprise metamodel, thus resulting in an ontology-enriched modelling language of

the enterprise repository.

Model Repository. A model repository is the central architectural component of the

platform responsible for storage and management of models. A repository is a generic

model storage system, always configured by a specific metamodel library (see Fig. 2,

M1 Layer). The main core elements of the model repository are: repository, reposi-

tory instance, model, modelling instance, attribute instance, relation instance, end-

point instance etc. Each repository concept has its corresponding concept in the

metamodel library so that the “conformsTo" relationship between M1 and M2 in-

stances is fulfilled. Unlike conventional repositories, having the integrated ontology-

enriched metamodel, the model repository is able to manage ontologies as enrichment

to models. Usually, only semantics of the models not expressible in conventional

languages are stored in ontologies such as different kinds of constraints.

Ontology Repository. An ontology repository serves as a global storage of ontolo-

gies. Various domain ontologies and upper-level ontologies are managed within this

repository. For example, reference domain ontologies are stored here, which represent

formally described semantics of a particular domain, against which the corresponding

models from the model repository on M1 level are validated. The similar is true for

the modelling languages defined on the M2 layer in the metamodel library. Here,

modelling languages are validated against their corresponding reference, upper-level

ontologies.

2.2 Ontology-aware Model Mechanisms Building Block

The mechanisms building block represents an extensible set of various platform fea-

tures all working on the ontology-aware model management building block. Typi-

cally, mechanisms use meta-information stored either on M3 or on M2 level, in order

to manipulate or work on instances on M2 or on M1 level, respectively. In the follow-

ing, the main platform mechanisms are described in more detail.

Integrated Querying. Querying a model repository is a basic capability which makes

comprehensive model information available. In an ontology-enriched modelling envi-

ronment, both model query languages like GReQL [9] or OCL[10] as well as ontol-

ogy query languages like SPARQL[11] are needed, to support querying in an inte-

grated way. A proposed platform support for integrated querying may include an

evaluator of the integrated query language which transforms part of the queries to

either model QL or ontology QL (see Fig. 3). Such queries are then executed against

the model i.e. ontology repository. Query language transformation engine also per-

forms translation of the query results back to the integrated query result. An existing

approach for integrating querying is the TWOUSE OCL [8]. The TWOUSE OCL is

an extension of the OCL with support to built-in operations which calls reasoning

services to reason over models after they are translated into OWL.

Fig. 3. Integrated Querying Mechanism.

Metamodel Mapping. Metamodel mappings have an important role in the metamod-

elling environment by adding knowledge about integrative usage of different model-

ling languages, still leaving their metamodel independent. A metamodel mapping

represents a structural and semantic correspondence between concepts of two meta-

models. One mapping takes at least one metamodel element (e.g. class, attribute or

relationship) from each of the source metamodels and relates them, in order to denote

the nature of the appearing structural and semantic conflicts. Bridging of metamodel-

ling and ontology languages is done via mappings [12]. Mapping may build basis for

MDA-based model transformation rules [13] or even for metamodel composition

rules [14]. The role of ontologies here is to facilitate identification of mappings.

Metamodels are translated to ontologies in order to infer possible concept mappings.

Model Transformation. Model transformation is a backbone mechanism in model-

driven approaches, since transformation engines enable MDA model chains (CIM-

PIM-PSM). Business process model refinement is a typical scenario for model trans-

formations within enterprise repositories. Manipulating a model repository, transfor-

mations are also used for model migration scenarios. Model transformation is a func-

tion that receives a source model, a source metamodel, a target metamodel and a set

of rules as input and produces a target model which conforms to a target metamodel.

The model transformation approach reuses the MDA architecture and defines a trans-

formation rule language as a metamodel on the M2 layer of the architecture. The

concrete transformation rules conform to this grammar and define how models are to

be translated between different modelling languages [15]. Ontologies may improve

the generation of model transformation rules [16].

Model Merge, Diff, Copy and Global Change. Model management mechanisms

like merge, diff and copy are generic operators executed on models [17] in a model

repository. The merge operator takes two or more models as input and produces as

output the union of models as a new model. The diff operator computes the difference

between two models. The difference between models represents the set of model

elements (also as a model) in one model that do not correspond to any model ele-

ments in the other model. The operator copy takes one model as input and returns a

copy of that model. The returned model includes all of the relationships of the input

model, including those that connect its objects to objects outside the model. The

global change operator applies global changes to objects of a model repository or of a

particular fragment of a repository. The global change operator changes a property

value or a set of property values which represent common properties/attributes of the

selected set of the objects. For example, a global change operator may change the

names of all model elements in one model or in the whole model repository, so that

each model element name begins with the certain prefix. The target set of models the

global change operator should work on can be determined by querying the model

repository.

Model Versioning. The versioning mechanism allows management of different

model versions in model repository over time. The unit of versioning can be either

model or a definable set of model elements. The versioning mechanism utilizes opera-

tors like merge, copy and diff.

Model Validation. Validation mechanisms on models and metamodels are used, in

order to support correct modelling. Basically, one can perform syntax and semantic

checks on models. Syntax checks validate a model against its grammar, i.e. meta-

model, to find out whether the models satisfy all the relevant well-formedness rules.

On the other side, models can be also checked for semantic consistency. Semantic

100

checks validate models against defined semantic constraints. The prerequisite for

automated semantic consistency checking are formally defined model semantics. A

model validator requires as input a model to validate which resides in the model re-

pository with the reference to its metamodel. A syntax checker validates a model for

its conformance to its metamodel. Semantic checks are done by a semantic checker.

The semantic checker translates the model part to the ontology and triggers a reasoner

to perform a semantic check.

Fig. 4. Model Validation.

Traceability Management. One of the challenges in modelling environments is

tracing of interdependencies between models and model elements created during the

course of modelling or during the execution of various mechanisms like, model

merge, transformation etc. For example, change impact analyses of enterprise models

which trace inter-level and intra-level dependencies of model elements may be real-

ized using traceability links. The solution may base on the similar approach which

analyzes impact changes in the code back to their requirements [18]. Traceability

relationships are created and stored within the model repository to track different

kinds of such dependencies. Each TR conforms to particular TR type which is part of

the metamodel and defines a set of possible relations between entities of different

metamodels. Traceability manager as a software component offers API-s to create

and query traceability links. Different platform mechanisms use then the traceability

manager to trace their operations. Traceability manager may also profit from a seman-

tic reasoner, for example, in inferring traceability links which are not explicitly mod-

elled in the repository.

Guidance. A guidance engine is a platform mechanism which guides the modeller in

achieving his goal within the particular modelling methodology. It is driven by the

guidance knowledge base. The guidance knowledge base consists of several informa-

tion sources. This may include formally described modelling process information, but

may also consist of existing modelling artefacts and/or interpretation of traceability

links. The architecture of the envisioned guidance mechanism is illustrated in Fig. 5.

Semantic Reasoning. The core mechanism which enables any ontology-aware ap-

proach is a semantic reasoner. Semantic reasoner offers a set of mechanisms which

enable inferencing over inference rules represented using e.g. a formal ontology lan-

guage. It is thus able to process queries on ontologies. A core building block is an

inference engine which receives, for instance, an OWL ontology as input, applies

inference production rules, and gives a query answer as output. A prominent imple-

mentation of the semantic reasoner is KAON2 [19].

101

Fig. 5. The Main Building Block of Guidance Mechanism.

2.3 Platform Access Services Building Block

Access services represent a building block which acts as a facade to the middleware

building blocks like ontology-aware model management and ontology-aware, model-

aware mechanisms (see Fig 1). On the one hand, the access services building block

enables the open, bi-directional exchange of all metamodel, model and ontology in-

formation. This is done by using various import/export services on model and meta-

model level. For example, depending on the core meta-metamodel implementation of

the platform, import and export interfaces to other meta-metamodels can be provided

such as KM3 [4], Ecore [3] or XMI. Further, access services building block covers all

aspects concerning security such as access rights, authorization, and en-/decryption.

API-s for scripting languages and other programming languages are also part of this

building block, which allows internal middleware components or UI building blocks

as well as external systems to access the metamodel library, the model repository and

mechanisms functionalities. Similarly, web service facade offers some of the platform

functionality as modular web services to other systems.

2.4 Model, Ontology and Mechanism Editors Building Block

Model, metamodel, ontology and mechanism editors form the building block repre

senting the presentation layer of the overall system (see Fig 1). Each of the editors

provides user interfaces for corresponding building block of the middleware layer and

is built according to a Model-View-Controller (MVC) architectural pattern. Since an

ontology-ware modelling environment is dedicated to ontology-aware model-

intensive tasks, a rich graphical editor for ontology-enriched model design is a man-

datory feature. The design of such editor therefore leverages advantages of the inte-

grated modelling language by supporting creation, editing, querying of combined

models. Additionally, it also allows for decoupled, independent views on models on

the one side and ontologies on the other side.

3 Summary and Future Work

In this research-in-progress paper we elaborated on the architecture of the ontology-

aware metamodelling platforms as environments for realizing advanced enterprise

102

repositories. The main contribution of the work is a generic architectural blueprint for

building such systems. An in-depth view of the main building blocks was given, fo-

cusing on how ontologies and semantic technology may be used to enhance the plat-

form. At the same time, this paper presents the preliminary results of the work done

within the EU funded FP7 research project MOST. The project MOST will improve

software engineering by leveraging ontology and reasoning technology. To reach this

goal, it aims at developing a seamless integration technology for ontologies into

model-driven software development, resulting in ontology-driven software develop-

ment (ODSD). Compared to the conventional approaches, ODSD should provide

better understanding of the software artefacts, in particular of models, by applying the

reasoning mechanism. Furthermore, it should provide better understanding of the

relationships between artefacts in different phases of the software development proc-

ess via sophisticated traceability techniques. Finally, ODSD should lead to improved

understanding of the software development process itself by applying ontology-

driven guidance. Within the project, we use the ADOxx metamodelling platform as a

modelling and a metamodelling environment which should offer advanced ontology-

aware model repository and guidance capability.

We are currently working on the first prototype evaluation of this architecture by

implementing the integrated metamodel and by integrating various ontology-aware

and model-aware mechanisms into the platform.

Acknowledgements

This research has been co-funded by the European Commission and by the Swiss

Federal Office for Education and Science within the 7

Framework Programme

project MOST N° 216691, http://most-project.eu.

References

1. Karagiannis, D., Kühn, H.: Metamodelling platforms, Invited paper.: In Proceedings of the

Third International Conference EC-Web 2002 – Dexa 2002. Springer-Verlag, Berlin, Hei-

delberg (2002)

2. Zivkovic, S., Murzek, M., and Kuehn, H.: Bringing Ontology Awareness into Model

Driven Engineering Platforms. In Proceedings of the 1st International Workshop on Trans-

forming and Weaving Ontologies in Model Driven Engineering (TWOMDE 2008), pages

47-54. http://sunsite.informatik.rwthaachen.de/Publications/CEUR-WS/Vol-395/. (2008)

3. Meta Object Facility (MOF) Specification-V2.0. http://www.omg.org/docs/ptc/03-10-

04.pdf

4. Steinberg, D., Budinsky, F., Paternostro, M., and Merks, E..: EMF: Eclipse Modeling

Framework 2nd Edition, Addison-Wesley Professional, Eclipse Series. (2008)

5. Jouault, F. and Bezivin, J.. KM3: A DSL for Metamodel Specification: In Proceedings of

8th IFIP International Conference on Formal Methods for Open Object-Based Distributed

Systems, pages 171-185. LNCS. (2006)

6. ATLAS. Atlas Megamodel Management (AM3). http://www.eclipse.org/gmt/am3/ (2006)

103

7. Ontology Definition Metamodel (ODM) – Version 1.0, Beta 3,

http://www.omg.org/spec/ODM/1.0/Beta3/PDF/ (2008)

8. Silva Parreiras, F., Staab, S., Winter, A.: TwoUse: Integrating UML Models and OWL

Ontologies. Universität Koblenz-Landau, Fachbereich Informatik. Nr. 16/2007. Arbeits-

berichte aus dem Fachbereich Informatik. Koblenz (2007)

9. Kullbach, B., Winter, A.: Querying as an Enabling Technology in Software Reengineering.

In: Verhoef, C. (Hrsg.); Nesi, P. (Hrsg.): Proc. of the 3rd Euromicro Conference on Soft-

ware Maintenance & Reengineering. Los Alamitos: IEEE Computer Society, (1999), S.

42–50

10. Object Constraint Language (OCL) Specification, Version 2.0, http://www.omg.org/cgi-

bin/doc?formal/2006-05-01 (2006)

11. SPARQL Query Language for RDF, W3C Reccomendation, http://www.w3.org/TR/rdf-

sparql-query/ (2008)

12. Silva Parreiras. F., Staab, S., and Winter, A.: On marrying ontological and metamodeling

technical spaces. In Proceedings of the 6th joint meeting of the European Software Engi-

neering Conference and the ACM SIGSOFT International Symposium on Foundations of

Software Engineering, 2007, Dubrovnik, Croatia, September 3-7. (2007)

13. Lopes, D., Hammoudi, S., Bezivin, J., and Jouault, F.: Mapping specification in MDA:

From theory to practice. In: Konstantas, D., Bourrieres, J.-P., Leonard, M., Boudjlida, N.

(Eds). Interoperability of Enterprise Software and Applications - INTEROP-ESA, p.253-

264. (2006)

14. Zivkovic, S., Kuehn, H., and Karagiannis, D.: Facilitate modelling using method integra-

tion: An approach using mappings and integration rules. In: Oesterle, H.; Schelp, J.; Win-

ter, R. (Eds.): Proceedings of the 15th European Conference on Information Systems

(ECIS2007) - "Relevant rigour - Rigorous relevance", St.Gallen, Switzerland, (2007) pages

2038-2050. http://is2.lse.ac.uk/asp/aspecis/20070196.pdf

15. Atlas Transformation Language (ATL). http://www.eclipse.org/m2m/atl/ (2008)

16. Roser, S. and Bauer, B.: An approach to automatically generated model transformations

using ontology engineering space. In 2nd International Workshop on Semantic Web En-

abled Software Engineering (SWESE), (2006)

17. Bernstein, P. A.: Applying model management to classical meta data problems. In Proceed-

ings of CIDR Conference 2003, Asilomar, CA, pages 209-220. (2003)

18. Schwarz, H., Ebert, J., Riediger, V., Winter, A.: Towards Querying of Traceability Infor-

mation in the Context of Software Evolution. In: GI Proceedings Bd. 126, (2008)

19. KAON Tool Suite Home Page, http://kaon.semanticweb.org/ (2009)

104