A METAMODEL INTEGRATION FOR METRICS

AND PROCESSES CORRELATION

Xabier Larrucea and Eider Iturbe

European Software Institute Tecnalia, Parque tecnologico de Zamudio 204, 48170, Zamudio, Spain

Keywords: Processes and metrics definition and execution, Model driven engineering.

Abstract: Nowadays organizations need to improve their efficiency mainly due to the current economic situation.

Several organizations are involved in process improvement initiatives in order to become more competitive

in the market. These initiatives require processes definition and performance measurement activities. This

paper describes briefly a metamodel integration between metrics metamodel and software and business

execution metamodels in order to support this kind of improvement initiatives. In fact this integration

implies to control coherently Software Metrics Metamodel for metrics, Software Process Engineering

Metamodel 2.0 for defining processes and JBPM Process Definition Language for executing processes. This

approach is supported by a prototype.

1 INTRODUCTION

Nowadays organizations need to improve their

efficiency due to an increase of global competency

in their markets. In fact they are starting up some

improvement activities related to quality models.

Specifically in the area of software quality and

software development projects there is a great

interest on process management. In fact this is

related to one of the eight elements of software

quality management (Humphrey, 2008): establish

and maintain statistical control of the software

engineering process. In this sense and particularly in

the software quality area we need not only define

and execute software processes but also to provide a

statistical control over them. Software and business

processes are intertwined in many organizations but

all of them are considered the organizations’

engines. In this context metrics are the basis for a

statistical process control and they are used to

measure organization’s behavior and performance

(Florac et al., 1999). Several seminal papers have

been published in the context of statistical process

control (Oakland 2007), (Card, 1994) applied to

software engineering.

Most of business process metrics are used to

calculate the complexity, coupling, cohesion, size

and modularity of business processes based on the

elements used such as in (Mendling, 2009),

(Vanderfeesten, 2007) and (Cardoso, 2007).

Therefore business process metrics have been

studied in for years and several metrics models has

been applied to analyze their complexity. This is a

similar situation with respect to software process

metrics models where (Genero et al., 2005), (Garcia,

2004) and (Rolón Aguilar, 2006) are studies

highlighting some metrics for software process

models.

In the context of metamodelling a huge effort has

been invested to develop metamodels for a wide

range of domains. In terms of metamodelling our

approach is compliant to the four architectural

metalayers promoted by the Object Management

Group (OMG). In our context some recent efforts

have produced the Software Process Engineering

Metamodel2.0 (SPEM, 2008) at OMG for describing

software processes. In addition in the metrics

knowledge area Software Metrics Metamodel

(SMM, 2009) is developed for defining metrics.

Moreover in the area of execution languages

there are several languages such as JBPM Process

Definition Language (JPDL) and Business Process

Execution Language (BPEL). Our approach is to

combine all these metamodels in order to provide a

coherent way to define and to model processes and

metrics and afterwards to provide control over the

execution of processes and their related metrics.

This paper is structured as follows. Firstly an

overview of current approaches is highlighted.

Secondly an integration amongst metamodels is

described. Thirdly an example of this approach is

Larrucea X. and Iturbe E. (2010).

A METAMODEL INTEGRATION FOR METRICS AND PROCESSES CORRELATION.

In Proceedings of the 5th International Conference on Software and Data Technologies, pages 63-68

DOI: 10.5220/0003039300630068

 SciTePress

MOF

SPEM2.0

JPDL

SMM

Instantiation

Metrics

Catalogue

Organisation’s

processes

Execution

Figure 1: Model Driven Architecture Metamodels approach.

shown. And finally a conclusion section is provided

in order to summarise the approach.

2 CURRENT APPROACHES

On the one hand metrics in software engineering has

been widely studied such as in (Fenton et al, 1998)

and (Kan 2004) where some metrics approaches are

underpinned. (Ng Keng Yap et al. 2008) provide a

robust data model for representing metrics.

On the other hand and as I mentioned before there

are some approaches such as (Mendling, 2009)

where author explores process models metrics such

as complexity, coupling, cohesion, size and

modularity. This approach is mainly focused on

process models and their properties. Our approach is

more similar to (Piattini et al. 2002), (Genero, 2005)

and (Chatters et al. 1998) where the focus of the

software quality is on measurements and metrics of

processes performance. (Chatters et al. 1998)

provide an example over a Cellular Manufacturing

Process Model using predicting models for costs.

(Piattini et al. 2002) outlines some steps for metrics

definition and validation. In fact with respect to

validation, authors identify theoretical validation and

empirical validation. Our approach aims to

overcome both approaches taken into account

measurement theory-based approaches and its

performance in empirical and real situations.

But there are more metrics applied to different

aspect such as in (Genero, 2005) where the authors

provide an overview of the existing proposals of

metrics for conceptual metrics. These related works

contribute to the set of metrics used in our approach.

In fact our architecture is based on “the process

virtual machine” provided by (Baeyens et al. 2010).

3 INTEGRATION METAMODELS

Our approach is basically based on the four-layer

metamodelling architecture promoted by the OMG

and its Model Driven Architecture.

Figure 1 summarises this approach where the highest

metamodel is MOF (Meta Object Facility) (layer

M3) and it is the basis for describing the subsequent

metamodels through an instantiation process. Our

proposal is to integrate SPEM2.0, SMM and JPDL

metamodels coherently from a metamodelling

perspective (layer M2).

Some parts of the used metamodels are not

implemented because they are out of our scope.

Figure 2 describes SMM metamodel elements

implemented in our approach. We have extended

this metamodel with the “variable” concept used for

implementing and using different types for metrics

measurements. Therefore we are able to define

different types of measurements to processes and we

have connected them to SPEM2.0 metamodel and

JPDL metamodel (Figure 4). In addition

“measurement” concept is related to “Element” that

is the connection point with SPEM2.0 metamodel.

Figure 3 represents a snapshot of the SPEM2.0

metamodel where the main elements are “Process”

and “TaskUse” representing processes and tasks

elements used for the relationship with SMM. These

elements are the cornerstone for this

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

Figure 2: SMM metamodel elements.

Figure 3: SPEM2.0 metamodel elements used in our approach.

integration because during definition and execution

of metrics and processes we maintain a correlation

between them.

As SPEM2.0 reuses UML activity diagram

concepts for flow definitions, we have selected

“fork”, “join” and “decision” concepts as part of the

behavior diagrams. All these elements are

represented in JPDL concepts (Figure 4). This JPDL

metamodel is only used for the final instantiation

process (see Figure 1). This final step corresponds to

A METAMODEL INTEGRATION FOR METRICS AND PROCESSES CORRELATION

Figure 4: JPDL metamodel extracted from current XSD.

Figure 5: A scrum process modeled using EPF.

the process runtime and to its execution. In fact it is

during process runtime when the correlation

between metrics and processes definition and their

execution should be maintained.

This integration provides us the ability to

instantiate processes for an organization and to

define a metrics catalogue that can be applied to

these processes. Moreover the resulting approach

allows us to instantiate during application runtime

user’s defined metrics depending on user’s

requirements. These users’ metrics can be linked to

processes and tasks.

Once processes and metrics are defined there is a

transformation between SPEM metamodel and JPDL

metamodel in order to deploy organizations’

processes into the resulting application.

4 USE CASE

We have applied our approach to a software

development process based on a Scrum

methodology/framework (Schwaber, 2002) for an

own development.

Scrum is an agile process or framework for

managing agile projects as described in

[http://www.scrumforteamsystem.com/ProcessGuida

nce/v2/FAQ/FAQ.aspx#whatIsScrum]. In this

context there are three main roles involved in a

scrum project: Product owner, Sprint Master and

Sprint team. The work to be done on a Scrum project

is listed in the Product Backlog, which is a list of

functional and non-functional requirements

including estimation of efforts in order to turn each

into deliverable product increments. In Scrum,

projects progress via a series of month-long

iterations called sprints. At the beginning of each

sprint, the scrum team selects a subset of

functionalities from the product backlog and they

work on it during the sprint, synchronizing their

tasks in daily scrum meetings.

In a scrum project there is a preparation phase called

sprint 0 and it is performed before a sprint iterations

cycle, where an initial prioritized product backlog is

defined and an initial release plan is created.

Based on this context we have used SPEM2.0

and the Eclipse Process Framework to define this

Scrum process (see Figure 5). In fact there are

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

Figure 6: Process Factory interface during import.

Figure 7: Adding a custom metric to a process.

several processes described using SPEM2.0 but it is

not the scope of the paper to describe all these

processes. Once we have defined this scrum process,

we import the process and we transform a SPEM2.0

instance to a JPDL instance. The relationship

between these elements is maintained through a

database. At the same time we can assign metrics to

this process from a metrics’ catalogue or from a user

definition. Some snapshots are provided in Annex

(see Figure 6 and Figure 7). Thus we maintain

integration not only at conceptual level but also at

implementation level.

5 CONCLUSIONS

This position paper provides an approach to

integrate processes and metrics from definition and

execution points of view. This integration is not only

at conceptual level that means we have extended all

these metamodels and we have related them

coherently. But also we provide a platform for

definition and execution for metrics and processes

where the connector elements maintain the

correlation amongst them.

As future work we are planning to substitute

A METAMODEL INTEGRATION FOR METRICS AND PROCESSES CORRELATION

EPF tool by SAP gravity

(http://www.sapweb20.com/) a web based tool for

process modelling and some extension to open

source tools such as ERPs.

We are working on a cloud based architecture

supporting the full life cycle of this approach. In this

sense we are testing our prototype in Google App

Engine (GAE) (http://code.google.com/intl/

en/appengine/).

ACKNOWLEDGEMENTS

This work has been supported by Vulcano project

TSI-020301-2009-1 funded by the Spanish

government (Ministerio de Industria, Turismo y

Comercio) and FACIT-SME (FP7-2008-2-contract

number: 243 695)

REFERENCES

Florac W., Carleton, A. “Measuring the software process:

statistical process control for software process

improvement”. Addison-Wesley Longman Publishing

Co., Inc. Boston, MA, USA Year of Publication: 1999

ISBN:0-201-60444-2

Oakland J.S. Statistical Process Control. Elsevier. ISBN-

13: 978-0-7506-6962-7

Card D. “Statistical Process Control for Software?” IEEE

Software Volume 11, Issue 3 (May 1994) Pages: 95 –

97. Year of Publication: 1994 ISSN:0740-7459

Mendling J. Metrics for Process Models. LNBIP 6. ISBN

978-3-540-89223-6. Springer Berlin Heidelberg, 2009,

p. 103

Vanderfeesten I., Cardoso J, Mendling J, Reijers H. , Wil

van der Alst. “Quality Metrics for Business Process

Models”. In L. Fischer, editor, BPM and Workflow

Handbook 2007, pages 179-190. Future Strategies

Inc., Lighthouse Point, Florida, USA, 2007.

Cardoso, J. “Business Process Quality Metrics: Log-Based

Complexity of Workflow Patterns”. On the Move to

Meaningful Internet Systems 2007: CoopIS, DOA,

ODBASE, GADA, and IS LNCS Book ISBN : 978-3-

540-76846-3. Volume 4803/2010

Genero M., Piattini M., Calero C. “Metrics for software

conceptual models”. Jan 2005 ISBN: 9781860946066

García F, Ruiz F, Piattini M. “Definition and Empirical

Validation of Metrics for Software Process Models”.

PROFES 2004: 146-158

Rolón Aguilar E, Ruiz F, García F, Piattini M. “Applying

Software Metrics to evaluate Business Process

Models”. CLEI Electron. J. (2006)

Baeyens T, Valdes Faura M. The Process Virtual

Machine. http://docs.jboss.com/jbpm/pvm/article/ (last

view 8th March 2010)

Software Metrics Metamodel. Object Management Group.

http://www.omg.org/spec/SMM/1.0/Beta1/PDF/March

2009.

Software Process Engineering Metamodel 2.0. Object

Management Group- http://www.omg.org/spec/

SPEM/2.0/PDF/ 2008

Schwaber K, Beedle M. “Agile Software Development

with Scrum”.Prentice Hall 2002. ISBN 0-13-067634-9

Ng Keng Yap, Abdul Azim Abdul Ghani, Ali Mamat,

Hazura Zulzalil, The Robust Software Metric Data

Model Defined in XML”, IJCSNS International

Journal of Computer Science and Network Security,

VOL.8 No.2, February 2008

Chatters, B, Hendersson P.Rostron C., “SIMMER:

Software and Systems integration modeling metrics

and risks” EuroSPI 98, Goteborg, November 1998

Piattini M , Genero M, Calero C. “Data Model Metrics”,

Handbook of Software Engineering and Knowledge

Engineering Vol II. 2002 ISBN: 981-02-4974-8

Kan Stephen H. “Metrics and Models in software quality

engineering”. Addison Wesley. July 2004. Second

edition. ISBN: 02019156.

Fenton N. Pfleeger S. L. Software Metrics: A Rigorous

and Practical Approach, 2nd edition. 1998

ISBN:0534954251 PWS Publishing Co

ICSOFT 2010 - 5th International Conference on Software and Data Technologies