An Approach of Ontology Oriented SPEM Models

Validation

Miroslav Líška

Faculty of Informatics and Information Technologies, Slovak University of Technology

Ilkovičova 3, Bratislava, Slovakia

Abstract. SPEM 2.0 is used to define software and systems development

processes and their components. It separates reusable development knowledge

base from its application in processes. In this work we present a knowledge

oriented approach that supports design of SPEM models with respect of SPEM

architecture. The main idea is based on creation three ontologies that are a

SPEM ontology, a development knowledge ontology and an ontology of devel-

opment knowledge application. Ontologies are generated with XSL transforma-

tions from UML models and also from SPEM CMOF specification and they are

imported into the Protégé for reasoning purposes. At the conclusion paper dis-

cusses a future work to increase contributions of this approach.

1 Introduction

This paper presents our latest research in application of the Semantic Web technolo-

gies in the domain of software engineering based on the Model Driven Architecture.

We have chosen the Web Ontology Language as the Semantic Web technology and

the Software Process Engineering Meta-Model as the software engineering technolo-

gy. OWL is a language for defining and instantiating Web ontologies [1]. SPEM 2.0

is used to define software and systems development processes and their components

[2].

Our general goal was to link together these two technologies. OWL is standard of

the World Wide Web Consortium that is the main international standards organiza-

tion for the World Wide Web. SPEM is standard of the Object Management Group

that is leading not-for-profit standards consortium that produces and maintains com-

puter industry specifications based on the Model Driven Architecture. So it is possi-

ble to assume that linking of these technologies could be perspective.

Our concrete goal of this research was to create technique that supports validation

of used company’s development knowledge base in its real software processes. In

order to accomplish these goals we had to ontologize SPEM, a development know-

ledge base and also its configuration in a process.

Líška M. (2009).

An Approach of Ontology Oriented SPEM Models Validation.

In Proceedings of the 1st International Workshop on Future Trends of Model-Driven Development, pages 40-43

DOI: 10.5220/0002176000400043

 SciTePress

1.1 MDA and the Semantic Web

At present, it is missing a straightforward linking between OWL and SPEM. But this

fact does not mean that OWL (Semantic Web) and SPEM (Model Driven Architec-

ture) are not related. First they were connected through RDF and UML [3, 4]. One

issue that has been addressed was the problem that RDF properties are first class

entities and they are not defined relative to a class [5]. This difference has also been

transferred into OWL and UML difference [6]. Moreover RDFS as a schema layer

language has a non-standard and non-fixed-layer meta-modeling architecture, which

makes some elements in model, have dual roles in the RDFS specification [7]. Today

the main bridge that connects the Semantic Web with the Model Driven Architecture

is stated in the Ontology Definition Meta-Model [8]. ODM defines the OWL Meta-

Model specified in MOF (MOF – OWL mapping) and also the UML Profile for On-

tology modeling (UML – OWL mapping). This architecture can be extended with

additional mappings between the UML Profile for OWL and other UML Profiles for

custom domains. This creates a bridge from the Semantic Web into custom domain of

the Model Driven Architecture [9]. In our approach we have also used this kind of

bridge to connect SPEM with the Semantic Web. We have generated a SPEM ontol-

ogy from the SPEM meta-model (specified in CMOF) that has stated our required

mapping between SPEM and UML.

2 Solution

The SPEM 2.0 Meta-Model is a MOF-based model that reuses UML 2 Infrastructure

library (e.g. Class, Association) [10]. Its own extended elements are structured into

seven main meta-model packages. These are Core, Process Structure, Process Beha-

vior, Managed Content, Method Content, Process with Methods and Method Plugin

package. In order to achieve our defined goals we have worked with Method Content

package and Process with Methods package. The Method Content meta-model pack-

age provides the concepts for SPEM 2.0 users and organizations to build up a devel-

opment knowledge base that is independent of any specific processes and develop-

ment projects. The Process with Methods meta-model package defines new and rede-

fines existing structures for integrating processes defined with Process Structure me-

ta-model package concepts with instances of Method Content meta-model package

concepts.

At first, we have generated (using XSL) a SPEM Ontology from SPEM CMOF

specification in order to obtain all SPEM elements with their relationship. Moreover

it was also necessary to extend the SPEM Ontology with SPEM UML2 profile in

order to relate SPEM elements with their UML representations. Note this has created

possibility to reason whether created SPEM UML models are correctly represented

with the SPEM UML Profile. At next it was necessary to create UML Model that was

representing a company’s development knowledge. We have focused on simple Task

Definition “Prioritize Use Cases” that is performed by “Project Manager”, “System

Analyst” and “Software Architect” Roles Definition. Mandatory output (Work Prod-

ucts Definition) from this task is “Use Case Model” and optional outputs from this

task are “Risk List” and “Software Architecture Document”. We have created this

model with Sparxsystems Enterprise Architect and transformed it (with XSL) into the

Company1 Method Content ontology. This ontology has become a simple develop-

ment knowledge base. If a new project is going to reuse “Prioritize Use Case” Task

Definition, its use must be according to this ontology. For example, “Prioritize Use

Cases” Task Use must have as a mandatory output “Use Case Model” Work Product

Use. Other outputs are not required, because they are only optional. So we have

created another UML model that represents use of Method Content in a process and

transformed it (with XSL) into the Company1 Method Content Use ontology. To

relate Method Content Use elements with Method Content elements it is necessary to

use contentTrace relationship as is depicted in Fig. 1. Note in our approach we use

contentTrace relationship equally as Instance relationship. For example, “Project

Manager” Role Definition in Method Content is a class, where “Project Manager”

Role Use is its instance.

3 Ontology Reasoning

At last we have imported the SPEM ontology, the Company1 Method Content Ontol-

ogy and Company1 Method Content Use Ontology into the Protégé. The Protégé is a

free, open source ontology editor and knowledge-base framework [11]. Note that it is

important to share the same namespace (spem2) in every ontologies in order to the

Protégé can properly recognize elements from the Company1 Method Content Use

Ontology as individuals of the Company1 Method Content ontology that is speciali-

zation of the SPEM Ontology.

PrioritizeUseCa ses

SoftwareArchitect

SystemAnalyst

Proj ectManager

UseCa seM odel

RiskLis t

Softw areArchitectureDocument

PrioritizeUse Cases

UseCa seM odel

SoftwareArchitect

Proj ectManager

SystemAnalyst

«contentTrace»

«perform s»

«performs»

«perform s»

«manda toryOutput»

«contentTrace»

«optionalOutput»

«mandatoryOutput»

«optionalOutput»

«perform s»

«performs»

Fig. 1. Method Content (left) and Method Content Use (right).

After that we have executed reasoning to find out whether these ontologies are

consistent. The Protégé resulted that ontologies are consistent what we have expected

because they were not contradictional. In order to prove that our approach is able to

disallow inconsistent use of the Method Content in a process, we have stated, that

mandatory output from “Prioritize Use Cases” Task Use is “Risk List” Work Product

Use instead of required “Use Case Model” Work Product Use. The Protégé has re-

sulted that this was inconsistent, because the object property mandatoryOutput re-

quires as a domain individuals of “Prioritize Use Cases” Task Definition and as range

it requires individuals of “Use Case Model” Work Product Definition. Note that

Work Product Definition “Risk List” and Work Product Definition “Use Case Mod-

el” must be disjointed classes.

4 Conclusions

In this paper we have presented our latest research based on knowledge engineering

in MDE. A lot of work must be done and it is fair to say, that at the moment we have

only started to bridge SPEM with UML. At present, our approach cannot be used

without limitations. For example, necessary condition that every Method Content and

Method Content Use model must have at least one responsible relationship, one per-

forms relationship and one mandatory output relationship must be satisfied. If not,

than the reasoner cannot find inconsistency what is required.

Acknowledgements

This work was partially supported by the Scientific Grant Agency of Slovak Repub-

lic, grant No. VG1/0508/09.

References

1. Smith, M.K., Welty, Ch., McGuinness, D.L.: OWL Web Ontology Language Guide. W3C

Recommendation, (2004)

2. OMG: Software and Systems Process Engineering Metamodel. formal/2008-04-01, (2008)

3. Manola, F., Miller, E.: RDF Primer. W3C Recommendation, (2004)

4. OMG: UML Superstructure. ptc/2008-05-05, (2008)

5. Cranefield, S.: Networked Knowledge Representation and Exchange using UML and RDF.

Journal of Digital Information, Volume 1 Issue 8, (2001)

6. OMG: OWL Full and UML 2.0 Compared. www.omg.org

7. Pan, J., Horrocks, I.: Metamodeling Architecture of Web Ontology Languages, In Pro-

ceedings of the First Semantic Web Working Symposium, Stanford, (2001)

8. OMG: Ontology Definition Metamodel 1.0. ptc/2008-09-07 (Beta 3), (2008)

9. Gaševic, D., Djuric, D., Devedžic, V.: MDA and Ontology Development. Springer, Berlin, (2006)

10. OMG: UML Infrastructure. ptc/2008-05-04, (2008)

11. Rector, P., Drummond, N., Horridge, M., Wang, H., Dameron, O.: Advanced Reasoning

with OWL. In 9th International Protégé Conference, Stanford, (2006)