SUPPORTING SOFTWARE PROCESS MEASUREMENT BY

USING METAMODELS

A DSL and a Framework

Beatriz Mora, Felix Garcia, Francisco Ruiz and Mario Piattini

Alarcos Research Group, Department of Computer Science, University of Castilla-La Mancha

Paseo de la Universidad 4, Ciudad Real, Spain

Keywords: Measurement, MDA, SMML.

Abstract: At present the objective of obtaining quality software products has led to the necessity of carrying out good

software processes management in which measurement is a fundamental factor. Due to the great diversity of

entities involved in software measurement, a consistent framework is necessary to integrate the different

entities in the measurement process. In this work a Software Measurement Framework (SMF) is presented

to measure any type of software entity. In this framework, any software entity in any domain could be

measured with a common Software Measurement metamodel and QVT transformations. Besides, we

present a Software Measurement Modelling Language (SMML) in order to define the measurement models

with take part in the measurement process. Furthermore an example which illustrates the framework’s

application to a concrete domain is furthermore shown.

1 INTRODUCTION

The current necessity of the software industry to

improve its competitiveness forces continuous

process improvement. This must be obtained

through successful process management (Florac,

Carleton et al., 2000). Measurement is an important

factor in the process life cycle due to the fact that it

controls issues and lacks during software

maintenance and development. In fact, measurement

has become a fundamental aspect of Software

Engineering (Fenton and Pfleeger, 1997). Software

Processes constitute the work base in a software

organization. Companies therefore wish to carry out

an effective and consistent software measurement

process to facilitate and promote continuous process

improvement. To do this, a discipline for data

analysis and measurement (Brown and Dennis,

2004), and measure definition, compilation and

analysis in the process, projects and software

products, is needed.

The great diversity in the kinds of entities which

are candidates for measurement in the context of the

software processes points to the importance of

providing the means through which to define

measurement models in companies in an integrated

and consistent way. This involves providing

companies with a suitable and consistent reference

for the definition of their software measurement

models along with the necessary technological

support to integrate the measurement of the different

kinds of entities. With the objective of satisfying the

exposed necessities, it is highly interesting to

consider the MDE (Model-Driven Engineering)

paradigm (Bézivin, Jouault et al., 2005) in which

software measurement models (SMM) are the

principal elements of the measurement process. Its

main goal is to ensure that the core artifacts in

software engineering processes will be models rather

than code. MDA (Model-Driven Architecture) is the

OMG proposal by which to carry out the MDE

Paradigm. The core of MDA is a set of standards

(MOF, QVT, OCL and XMI). According to the

QVT standard, the software development process is

a set of model transformations, from an abstract to a

specific level. The requirements are in the more

abstract level and the code is in the more specific

level.

Software measurement can benefit from the

MDE paradigm, providing integration and support to

carry out an automatic software measurement of any

software type. This implies that: a) the definition of

305

Mora B., Garcia F., Ruiz F. and Piattini M. (2008).

SUPPORTING SOFTWARE PROCESS MEASUREMENT BY USING METAMODELS - A DSL and a Framework.

In Proceedings of the Third International Conference on Software and Data Technologies, pages 305-312

DOI: 10.5220/0001897003050312

 SciTePress

measurement models conform to a Software

Measurement metamodel; b) the definition of

generic measurement methods are applicable to any

model-based software artifact; and c) support for

computing measures, for storing results and for

enhancing decision making.

The availability of a language which allows us to

represent those elements which must be taken into

account in the measurement processes might,

therefore, be important in decision making and in

process improvement. It is thus of interest to

consider the use of Domain Specific Languages

(DSLs) such as the Software Modelling

Measurement Language (SMML)(Mora, Ruiz et al.,

2008).

These aspects constitute the main interest of this

paper: in which the application of MDA principles,

standards and tools are used in software

measurement. We present the Software

Measurement Framework (SMF), a generic

framework to define measurement models which

conform to a common measurement metamodel, and

to measure any software entity with regard to a

domain metamodel. MOMENT environment has

been used, which supports the automatic model

management MDA compliant. These measurement

models involved in the framework can be defined by

using SMML. SMML is integrated in SMF and

permits software measurement models to be created

in a simple and intuitive manner. This language has

been done by using the Software Measurement

Metamodel (SMM)(García, Serrano et al., 2007),

(Mora, Ruiz et al., 2008) as the Domain Definition

Metamodel (DDMM). The task of the SMML is to

facilitate the definition of software measurement

models, which is the starting point in the generic

software measurement process.

The remainder of the paper is organized as

follows. Section 2 provides an overview of related

works. Section 4 presents the SMML and Section 3

describes the Software Measurement Framework

(SMF), including conceptual architecture,

technological aspects, and method. In Section 5 the

use of the framework and SMML is illustrated with

an example. Finally, conclusions and future works

are outlined in Section 6.

2 RELATED WORKS

We have found numerous publications which deal

with tools that have important success factors in

software measurement efforts (Komi-Sirviö,

Parviainen et al., 2001), which supply work

environments and general approximations

(Kempkens, Rösch et al., 2000), or which give

architectures more specific solutions (Jokikyyny and

Lassenius, 1999), (Brown and Dennis, 2004)

includes a list of tools which support the creation,

control and analysis of software measurements.

(Auer, Graser et al., 2003) furthermore examines

various software measurement tools, such as

MetricFlame, MetricCenter, Estimate Professional,

CostXPert and ProjectConsole, in heterogenic

environments.

It is also possible to find certain proposals

through which to tackle software measurement

which are more integrated and less specific than in

the aforementioned cases. (Palza, Fuhrman et al.,

2003) proposes the MMR tool which is based on the

CMMI model for the evolution of software

processes, and it is possible to consult similar tools

in (Harrison 2004), (Scotto, Sillitti et al., 2004),

(Lavazza and Agostini, 2005). These proposals are,

however, restricted to concrete domains or to

evaluation models of specific quality.

(Vépa, Bézivin et al., 2006) presents a metamodel

which allows the storage of measurement data, and a

set of transformations through which to carry out the

measurement of models based on a metamodel is

presented. This paper focuses upon the technological

aspects needed to implement the software

measurement with ATL technology, by offering the

user a variety of graphic representations of the

measurement results obtained.

This final proposal and that which is presented

here are complementary as they both focus upon two

key support elements of generic measurement: the

conceptual base, which is the main contribution of

FMESP, and technological implementation. Some

differences from technological point of view exist.

The measurements which are applied in the work

of (Vépa, Bézivin et al., 2006) are previously

defined in the ATL transformation archives. The

measurable entities are typical of the metamodels

presented in this work (KM3 (Jouault and Bézivin,

2006) and UML2). For example, the measurable

entities for a model which is expressed in km3 might

be package, class, attribute, reference etc.

The measurements in the proposal presented here are

defined by the user, i.e. the model transformation

needed to carry out the measurement it is not a

model previously defined, but this model is defined

according to the users needs. The measurement

definition is possible thanks to the software

measurement model, which contains all that is

relative to the measurement to be carried out in each

case. Moreover, the measurable entities are those

ICSOFT 2008 - International Conference on Software and Data Technologies

306

Table 1: A selection of the SMML elements and icons.

Information need Entity Base Measure Scale

____________

Quality Model Attribute Derived Measure Unit

__________

Description Measurable Concept Indicator Analysis model

Analisys

model

which are defined in their corresponding domain and

measurement metamodel (expressed in ecore). A

further difference is that SMF uses QVT.

Finally, it is important to mention a method for

specifying models of software data sets in order to

capture the definitions and relationships among

software measures presented in (Kitchenham,

Hughes et al., 2001).

3 SOFTWARE MEASUREMENT

MODELLING LANGUAGE

(SMML)

SMML (Mora, Ruiz et al., 2008) is a language

which permits software measurement models to be

built in a simple and intuitive manner. The Software

Measurement Metamodel (SMM) which is derived

from the Software Measurement Ontology

(SMO)(García, Bertoa et al., 2006) defines the

abstract syntax of SMML. The Software

Measurement Metamodel supports the graphical

language to represent in an intuitive way software

measurement models.

With regard to expected requirements (Kolovos,

Paige et al., 2006), we shall now show the

requirements which are valid in our Language: a)

Conform: the language constructs correspond to

important domain concepts; b) Orthogonal: Each

language construct is used to represent exactly one

distinct concept (Attribute, Base Measure, etc.) in

the domain; c) Supportable: The SMML language is

supported by tools such as MS/DSL Tools or GMF

(Eclipse, 2007); d) Simple: the DSL is simple in

order to express the domain concepts and to support

its users; e) Usable: DSL constructs are expressive

and easy to understand. Table 1shows some of the

most representative graphical elements of the

SMML (for a greater detail see (Mora, Ruiz et al.,

2008)).

4 SOFTWARE MEASUREMENT

FRAMEWORK

The Software Measurement Framework (SMF) (for

greater detail see (Mora, García et al., 2008))

permits us to measure any type of software entity. In

this framework, any kind of software entity

represented by its corresponding metamodel in any

domain can be measured with a common Software

Measurement metamodel and QVT transformations.

SMF has three fundamental elements: conceptual

architecture, technological aspects and method.

SUPPORTING SOFTWARE PROCESS MEASUREMENT BY USING METAMODELS - A DSL and a Framework

307

These elements have all been adapted to the MDE

paradigm and to MDA technology, taking advantage

of their benefits within the field of software

measurement. The Software Measurement

Framework (SMF) is the evolution of the FMESP

(García, Piattini et al., 2006), but is adapted to the

MDE paradigm and uses MDA technology.

The following subsections explain briefly the

conceptual, technological and methodological

elements which are part of SMF.

4.1 Conceptual Architecture

Due to the necessity of having a generic and

homogeneous environment for software

measurement (García, Bertoa et al., 2006; García,

Piattini et al., 2006; García, Serrano et al., 2007), a

conceptual architecture and a tool with which to

integrate the software measurement are proposed. In

the following section, the main characteristics of this

proposal are described. In (García, Serrano et al.

2007) a more detailed description can be found.

MOF

Software

Measurement

Metamodel

Measurement

Models

Domain

Models

Data

Integrated Measurement: Conceptual Framework

Model

Level

Meta-Model

Level

Meta-Meta-Model

Level

Domain

Metamodels

Domain

Metamodels

Data

Level

Figure 1: Conceptual framework with which to manage

software measurement.

SMF is part of the FMESP framework (García,

Piattini et al.,2006). The FMESP framework permits

representing and managing software processes from

the perspectives of modelling and measurement. We

focus on the measurement support of the framework

whose elements are detailed according to the three

layers of abstraction of metadata that they belong to,

according to the MOF standard. As can be observed

Figure 1, the architecture has been organized into the

following conceptual levels of metadata: Meta-

Metamodel Level (M3), Metamodel Level (M2) and

Model Level (M1).

In order to establish and clarify the concepts and

relationships that are involved in the software

measurement domain before designing the

metamodel, an ontology for software measurement

was developed (García, Bertoa et al., 2006). The

measurement metamodel was derived by using the

concepts and relationships stated in the ontology as a

base. The Software Measurement metamodel (which

is integrated in SMF) is organized around four main

packages (for greater detail see the work of (García,

Bertoa et al., 2006)): Software Measurement

Characterization and Objectives, Software

Measures, Measurement Approaches and

Measurement Action.

4.2 Technological Aspects

In this section the technological aspects of SMF are

explained.

• Adaptation to MDA. In the Figure 2 necessary

elements for the FMESP adaptation to MDA are

presented according to MOF levels.

Figure 2: Elements of the FMESP adaptation in a MDA

context.

As can be observed in Figure 2, two new elements,

namely the QVT Relations model and metamodel,

have been added to adapt the conceptual architecture

illustrated in Figure 1 to MDA. The QVT Relations

Model is obtained automatically through a

transformation from a Measurement model. It

contains all the information necessary to carry out

the QVT transformation of the SMF proposal. Ecore

language has been selected because it is an

implementation of EMOF. EMOF is the part of the

MOF 2.0 specification that is used for defining

simple metamodels using simple concepts.

Software

Measurement

Metamodel

Domain

Metamodel (1)

conforms to

QVT Relations

Metamodel

conforms to

conforms to conforms to

Transformation

(4)

ECORE

Software Measurement

Model (target)

Software

Measurement

Metamodel (2)

QVT

Relations

Model

Domain

Model (3)

ICSOFT 2008 - International Conference on Software and Data Technologies

308

Figure 3: QVT Relations transformation model.

• QVT Relations Transformation. The QVT

Relations model is the transformation needed to

perform the measurement. In this transformation

two source models are involved: a Software

Measurement model and a domain model; the

target model is the Software Measurement Model

with the measurement results (see Fig. 2). Due to

the fact that the proposal is about generic

measurement, it is very important that the QVT

model is obtained in a generic way. The MDE

paradigm and MDA technology are applied for

this reason. This transformation is obtained

automatically from the previous QVT

transformation shown in Fig. 3. The QVT

Relations model, called the extended or final

QVT Relations model, is obtained from a QVT

transformation, where there are two source

models: the basic or initial QVT Relations model

(which conforms to the QVT Relations

metamodel) and the Software Measurement

model (previously defined). (for a greater detail

see (Mora, García et al., 2008)).

• Technological Environment. In this work, the

tool selected has been the model management

environment called MOMENT (MOment

manageMENT)(Boronat and Meseguer, 2007).

This framework is integrated in the Eclipse

platform. It provides a set of generic operators to

deal with models through the Eclipse Modelling

Framework (EMF)(2007). The underlying

formalism of the model management approach is

the algebraic language Maude.

4.3 Method

The necessary steps to carry out the software

measurement by using the SMF are explained below

(see Fig. 2):

1. Incorporation of Domain Metamodel: the

measurement is made in a specific domain. This

domain must be defined according to its

metamodel.

2. Creation of Measurement Model: the

measurement model is created according to the

Software Measurement metamodel which is

integrated in SMF. This first model is the source

model, so the results are therefore still not

defined, i.e. the “Measurement Action” package

from the Software Measurement metamodel is

still not instantiated. In order to facilitate the

measurement model definition, SMML can be

used.

3. Creation of Domain Model: which is defined

according to its corresponding domain metamodel

(created in the first step). The domain models are

the entities whose attributes are measured by

calculating the measurements defined in the

corresponding measurement models. Examples of

domain models are: the UML models (use cases,

class diagrams, etc.), or the E/R models.

4. Measurement Execution: the measurement

execution is carried out through QVT

transformation, in which, the measurement model

is obtained by starting from the two source

models (the measurement model and the domain

model) where the results are defined, i.e. the

“Measurement Action” package is instantiated.

The target measurement model is the extension of

the source measurement model. The measurement

results are calculated by running OCL queries on

the domain model.

5 CASE STUDY

To illustrate the benefits of the proposal, consider

the example of relational database measurement. For

greater simplicity, only the following elements are

shown in Fig. 4: Measurement Method, Entity (to

which the measurement method is applied) and

Measurement result (the result is obtained by

executing the measurement method on the entity).

Furthermore, it is necessary for the domain

metamodel, in this case Relational Databases

domain, to have been previously chosen. Both

metamodels are independent (see fig 4), although

they are logically related. In Fig. 4 the measurement

and domain metamodels have been represented in a

clear and a dark colour, respectively.

Basic QVT-Relations

Model (.qvt)

Extended QVT-Relations

Model (.qvt)

QVT Transformatio

Software Measurement

Model (target)

Measurement Execution

Transformation

Software Measurement

Model (source)

SUPPORTING SOFTWARE PROCESS MEASUREMENT BY USING METAMODELS - A DSL and a Framework

309

PrimaryKey

Foreign Key

Table

Attribute Key

<<abstract>>

Model Element

<<abstract>>

RelationalSchema

Measuremet Method

(from Measurement A pproaches)

Measurement Result

(from Meas urement Action)

Measurement

Entity Class

(from Characterization and Objectives)

Entity

(from Characterization and Objectives)

Figure 4: Relationship between Relational Database (domain) Metamodel and SMM.

__________

____________

Figure 5: Measurement Model with SMML.

In this example, the chosen measurement method

has been “COUNT elements of type TABLE”,

which is an instantiation of the abstract method

“COUNT elements of type X”.

In order to carry out the measurement, the following

steps (four steps) must take place:

1. Incorporation of Relational Databases metamodel

(represented in a dark colour in Fig. 4).

2. Creation of measurement model conforms to

Software Measurement metamodel. For the

measurement method “COUNT elements of type

TABLE”, the values of Entity and Measurement

Method are Table and Count, respectively. The

Measurement Result is not still defined. In this

case SMML can be used to define the

measurement model (see Fig. 5).

3. Creation of model conforms to the Relation

Database metamodel. In this case, the model

(relational schema) is a university domain

composed of five tables with their corresponding

primary keys, foreign keys, and attributes.

The extended QVT Relations model was needed

to carry out the fourth step. This transformation is

obtained automatically (see Fig. 6).

4. The source models used to carry out the

measurement are: the measurement model (2

step), the domain model (3

step) and the

extended QVT Relations model. The target model

obtained is the measurement model with defined

Measurement Result (Fig. 7). In this example the

value of Measurement Result is 5 (number of

tables).

In the same way as is illustrated with Relational

Databases, the method can be applied to any other

domains, such as for example, UML models, Project

Management or Business Processes, etc.

Figure 6: “Function” elements from extended QVT

Relations model.

Figure 7: Measurement result.

ICSOFT 2008 - International Conference on Software and Data Technologies

310

6 CONCLUSIONS AND FUTURE

WORK

In this paper, a generic framework for the definition

of measurement models based on a common

metamodel has been outline, and we have explained

how to work with it. The framework allows the

integrated management and measurement of a great

diversity of entities.

Following the MDA approach and starting from a

(universal) measurement metamodel, it is possible to

carry out the measurement of any domain by means

of QVT transformation, and this process (QVT

transformation) is completely transparent to the user.

With SMF, it is possible to measure any software

entity. The user task consists in selecting the

domain metamodel (the domain to be measured) and

defining the source models. The software metamodel

is integrated in the framework.

In order to facilitate the measurement models

definition, a Software Measurement Modelling

Language (SMML) is used to supply measurement

engineers with the definition of software

measurement models according to the proposed

metamodel.

Among related future works, one important work

is the realization of a plug-in based on Eclipse which

will supply the user with the data introduction and

the measurement process. This plug-in will enable

users to instantiate measurement models in an easy

and intuitive way. Other future work will be to align

our metamodel with the Software Metrics Meta-

Model (SMM) OMG proposal (OMG, 2007).

Finally, we shall apply SMF to real environments to

obtain further refinements and validation.

ACKNOWLEDGEMENTS

This work has been partially financed by the

following projects: INGENIO (Universidad

Pólitecnica de Valencia, PAC08-0154-9262) and

ESFINGE (Ministerio de Educación y Ciencia,

TIN2006-15175-C05-05).

REFERENCES

(2007). "Eclipse Modelling Framework (EMF) Main

Page." from http://www.eclipse.org/emf.

Auer, M., B. Graser, et al. (2003). A Survey on the Fitness

of Commercial Software Metric Tools for Service in

Heterogeneous Environments: Common Pitfalls Ninth

International Software Metrics Symposium.(Metrics

'03).

Bézivin, J., F. Jouault, et al. (2005). Principles, standards

and tools for model engineering. 10th IEEE

International Conference on Engineering of Complex

Computer Systems (ICECCS’2005).

Boronat, A. and J. Meseguer (2007). Algebraic Semantics

of EMOF/OCL Metamodels, CS Dept., University of

Illinois at Urbana-Champaign.

Brown, M. and G. Dennis (2004). "Measurement and

Analysis: What Can and Does Go Wrong?" 10th IEEE

International Symposium on Software Metrics

(METRICS'04): 131-138.

Eclipse. (2007). "Eclipse Graphical Modeling Framework

(GMF) Main Page." from http://www.eclipse.org/gmf/

Fenton, N. and S. L. Pfleeger (1997). Software Metrics: A

Rigorous & Practical Approach, Second Edition, PWS

Publishing Company.

Florac, W. A., A. D. Carleton, et al. (2000). "Statistical

Process Control: Analyzing a Space Shuttle Onboard

Software Process." IEEE Software 17 (4)(4).

García, F., M. F. Bertoa, et al. (2006). "Towards a

consistent terminology for software measurement."

Information and Software Technology 48(8): 631-644

García, F., M. Piattini, et al. (2006). "FMESP: Framework

for the modeling and evaluation of software

processes." Journal of Systems Architecture - Agile

Methodologies for Software Production 52(11):627-639

García, F., M. Serrano, et al. (2007). "Managing Software

Process Measurement: A Metamodel-Based

Approach." Information Sciences.

Harrison, W. (2004). "A flexible method for maintaining

software metrics data: a universal metrics repository."

Journal of Systems and Software 72(2): 225-234

Jokikyyny, T. and C. Lassenius (1999). Using the internet

to comunicate software metrics in a large organization.

Proceedings of GlobeCom'99.

Jouault, F. and J. Bézivin (2006). "KM3: a DSL for

Metamodel Specification."

Kempkens, R., P. Rösch, et al. (2000). Instrumenting

Measurement Programs with Tools. PROFES 2000,

Oulu, Finland.

Kitchenham, B., R. T. Hughes, et al. (2001). "Modeling

Software Measurement Data." IEEE Transactions on

Software Engineering 27(9): 788-804.

Kolovos, D. S., R. F. Paige, et al. (2006). Requirements

for Domain-Specific Languages. First ECOOP

Workshop on Domain-Specific Program Development

(ECOOP'06), Nantes, France.

Komi-Sirviö, S., P. Parviainen, et al. (2001). Measurement

Automation: Methodological Background and

Practical Solutions-A Multiple Case Study. Seventh

International Software Metrics Symposium

(METRICS'01), London.

Lavazza, L. and A. Agostini (2005). "Automated

Measurement of UML Models: an open toolset

approach." Object Technology 4(4): 115-134.

Mora, B., F. García, et al. (2008). Software Measurement

by using QVT Transformation in an MDA context.

ICEIS 2008 (In Press), Barcelona (Spain).

SUPPORTING SOFTWARE PROCESS MEASUREMENT BY USING METAMODELS - A DSL and a Framework

311

Mora, B., F. Ruiz, et al. (2008). SMML: Software

Measurement Modeling Language, Department of

Computer Science. University of Castilla - La

Mancha.

OMG (2007). Architecture-Driven Modernization (ADM):

Software Metrics Meta-Model (SMM). OMG

Document: dmtf/2007-03-02, Object Management

Group.

Palza, E., C. Fuhrman, et al. (2003). Establishing a

Generic and Multidimensional Measurement

Repository in CMMI context 28th Annual NASA

Goddard Software Engineering Workshop (SEW'03),

Greenbelt (Maryland, USA).

Scotto, M., A. Sillitti, et al. (2004). A relational approach

to software metrics. Proceedings of the 2004 ACM

symposium on Applied computing (SAC'2004),

Nicosia, Cyprus.

Vépa, É., J. Bézivin, et al. (2006). Measuring Model

Repositories. Model Size Metrics Workshop at the

MoDELS/UML 2006 conference, Genoava, Italy.

ICSOFT 2008 - International Conference on Software and Data Technologies

312