ion and granularity levels. The enactment of such methods in projects entails risks that threaten its success. For instance, the set of modelling languages may have redundant and/or missing constructs, the modellers can attempt to use a language that is inappropriate for a given level of abstraction, if a language covers several abstraction levels then some of its constructs may have a semantic overload. This work proposes a framework to evaluate the ontological quality of a set of languages; that is, measuring to what extent the set of languages comply with principles and guidelines intended to minimise the above-mentioned risks and to facilitate their combined use within an MDE project. Also, the framework will allow to identify relations between language constructs and software concepts, so it is expected to aid MDE method engineers in the task of designing model transformations.
One of the main challenges in the model-driven engineering (MDE) initiative
is the management and the integration of languages and models formulated
to support multiple views during the information systems (IS) development.
Modelling languages create and use models that represent materialized views
over concerns of an IS, according to rules de ned by viewpoints. In this way, it is
possible to mitigate the problems associated with the management of transversal
features of an information system [
Generally, the considerations addressed by the languages and models contemplate: i) business concerns; ii) non-functional features derived from quality attributes; iii) new paradigms for software construction (e.g., aspects, collaboration or requirements characterization); iv) functional and logical concerns.
Now a proliferation of languages is evident (with their abstract and concrete syntax and their semantics), and proposals that emerge with the purpose of managing speci c views or perspectives of an IS. There are proposals that de ne their wide set of symbols and concepts, and which have not been interesting by the academical, researchers and industrial communities. Also, there are proposals based on excessively stereotyped UML, which limit the expressiveness or meaning of the models to the stereotyped classes, and/or modi cations (or additions) of UML symbols. New UML based notations could not fully satisfy the meaningmeaningful relation associated with a speci c domain. Therefore, people who designed a notation of this style should be able to transmit the meaning of the concept to express.
This paper presents a proposal for de ning the foundations of an ontological evaluation framework to be applied over languages used in MDE projects, with the purpose of validating the quality of these elements in the management and technical implementation of an IS according with the views (stakeholders) involved, and features of the MDE itself. This paper is organized as follows: the Section 2 introduces the problem statement that promotes our idea. Section 3 presents the goals and the methodology of our research. Section 4 brie y presents an overview of our proposed solution; and nally, Section 5 reviews the state of the art about quality of models and modelling languages. 2
MDE proposes modelling languages as the new abstraction units, hence, the
introduction of a new language in an MDE enviroment should be as easy as
creating a new class in a Java project [
The adoption of the MDE approaches have guided the development of a large number of initiatives; although it emphasizes the use of models as primary artifacts of a software construction process, it causes a conceptual divergence in the support of speci c views and/or concerns belonging to an IS. This phenomenon is strengthened by the lack of (semantic) support o ered by UML or other traditional notations.
In spite of the development of metamodels, reference architectural
frameworks, and ontological frameworks, it has been recognized and widely reported
the inability for consistently modelling all related and inherent views in an IS
using a single metamodel or a single notation. In [
Due to the increasing collection of modelling languages and notations, several methods to assess the quality of modelling languages have been proposed. Some proposals provide guidelines for designing languages based on principles drawn from semiotics and cognitive theory (see the state of the art in Section 5). The rationale behind such proposals is that models are a means to express conceptions about some phenomena, to reason about such conceptions, and to communicate them to others.
Although these methods emphasize the importance of the relationship
between the concepts of the modeled concern with respect to the used notation, the
required e ort to formalize semantic de nitions become a high cognitive load
for those involved in an MDE process. Also, these frameworks do not consider
the most relevant features of the MDE itself into their formulation. It can be
explained as a natural consequence of the several (divergents) interpretations
of MDE that results from attempts for new notations and languages framed in
MDE without a rationale support (particular interpretations on MDE). There
are so many ways to adopt an MDE approach that it is not possible to establish
general conclusions about MDE itself [
Also, the identi ed guidelines and frameworks do not evaluate the quality of models from dimensions such as the mapping or translation between models (even models that belong to the same viewpoint of an IS), neither successful experiences originated from massive application of a modelling technique in a particular MDE environment. 3
The main goal of this work is to formulate a method for the evaluation of the ontological quality of a set of languages jointly used within an MDE project. This work aims to verify whether it is possible to generate a framework for the evaluation of languages, so that it can determinate how one language from the MDE viewpoint is structured. It means, if the language supports views, abstraction levels, integration capabilities, and if it is possible to generate full functional software from the language(s)/model(s) under review. The framework must indicate what is missing or what is not neccesary for using a language in an MDE environment correctly. 3.1
The research will focus on resolving the following questions: { (RQ1) What problems are evidenced in model-driven projects related to the selection of languages? { (RQ2) What is the set of concepts that are required to model when we are in a model-driven project? { (RQ3) When a set of modelling languages is selected to be used in combination in an MDE project, are there methods for evaluating the suitability such set of languages? { (RQ4) Propose a method for the evaluation of the ontological quality of a set of languages used jointly within a model-driven project. { (RQ5) What advantages/disadvantages are obtained by the application of the proposed method? T3.8. ANALYSIS OF RESULTS
T3.7. RESEARCH T3.6. DESIGN VALIDATION T3.5. RESEARCH DESIGN T3.5.1 Design experiment for analyze the applicability of ontological evaluation framework
T3.4. PROBLEM INVESTIGATION T3.4.1 Define research goals
EC1.
Propose a method for RC1.
RC2. evaluating the ontological Systematic review RESEARCH CYCLE quality of a set of languages about quality in
Assess the used in combination into a MDE application of the MDE approach applied over ontological quality an IS
framework RQ 4 T3. SPECIFICATION VALIDATION (RQ5) T3.1 Perform theorical comparisons with
alternative solutions T3.2 Perform lab demos T3.3 Perform constrating interviews
T2.1. PROBLEM INVESTIGATION T.2.1.1. Analyze the definitions about quality in
MDE and their impact in model-driven engineering context T2.2. RESEARCH DESIGN T2.2.1 Define goals,contributions, questions, strings and data sources for the systematic review T2.2.2.Define qualification criteria for
classification of search results T2.3. DESIGN VALIDATION
T2.4. RESEARCH
T2.5. ANALYSIS OF RESULTS T2.6. Perfom a literature review about ontological reference
frameworks for information systems T2.7. Identify and assess methods for evaluating the selection or convenience of a languages in a model driven engineering project (RQ3) T2.8. Design of a model of ful functional information system
(FF model) according with ontological frameworks T2.9. Define the dimensions of the ontological quality for
languages in model driven engineering scenarios T2.10. Design the taxonomical structure for ontologicial evaluation
of quality T2.11. Define the satisfiability criteria (rules) of languages
according to the framework T2.12. Design guidelines for applying the ontological quality
evaluation framework T5. IMPLEMENTATION EVALUATION T5.1. Lab demos T5.2. Usablity evaluation T5.3. Apply potential improvements
detected T4. IMPLEMENTATION T4.1. Ecore/EMF metamodels T4.2. Method implementation
T1. PROBLEM INVESTIGATION T1.1 Identify problems in model-driven enginering projects related
to the selection of languages (RQ1) T1.2. Identify the set of concepts required to be modeled
in a model-driven project (RQ2) T1.3. Perform a literature review for Identifying existent reports about
language problems in previous model-driven experiences T1.4. State of the art in ontological evaluation frameworks for
languages in MDE start
T2. SOLUTION SPECIFICATION
In a practical way, through the proposed evaluation framework, the language
designer or language engineer[
This work will use the Design Science guidelines proposed in [
Our engineering cycle starts with the identi cation of the involved stakeholders through survey-based research and ethnography research. For this case, we consider people involved in model-driven projects, such as languages users and method engineers. This identi cation includes the (further) expectations about the use of combination of languages for developing IS under modeldriven principles. To answer RQ1 we will perform a literature review. RQ2 will be answered by means of a literature review of the most representative IS ontological reference frameworks contrasting their conceptions about elements of an IS with MDE features. To respond RQ3, a literature review must be performed in order to identify similar evaluation frameworks in ontological levels. RQ4 will be answered by the design of the model of the ontological framework for the evaluation of languages in MDE context, with its respective use speci cation. RQ4 implies a systematic review about the concept of quality in model-driven engineering to identify the relevance and scope of our proposal with respect to representative trends of quality in MDE. To answer RQ5, we will perform theoretical comparisons and cognitive analysis in which we assess the e ort needed to apply the proposal, a lab demo in which we apply our proposal in a small but realistic case, and a controlled experiment in which we evaluate our proposal in a rigorous way. 4
The quality evaluation framework proposed in this research is conceived as a conceptual, methodology and technology tool for the evaluation of language proposals and models (as productions of languages), whose purpose is to assess one set of languages/models regarding its incorporation and adoption capabilities in a MDE enviroment. Also, this framework must establish the capacity of languages to support automation and software generation. Figure 2 exposes the initial version of a metamodel that conforms the quality evaluation framework. This metamodel will be updated according with the re nement of the quality term as a consecuence of advances produced by the introduction of the most applicable philosophical background.
Figure 3 presents a initial collection of metaclasses which considers some typical elements existing into a MDE environment, to be assesed through the application of the evaluation framework proposed in this work.
The existence of several languages in a IS model-driven project could derivate evidences about those languages that overlap and model IS aspects in a redundant way, or conversely, some of the IS aspects could not be covered by any language. Both situations supose a risk for MDE projects. We think it in uences in the adoption of model-driven methods and tools. Therefore, when the languages and tools are established accordly, it will favor the adoption of modeldriven intiatives. When the framework is used, it will be possible to optimize the selection of languages; and therefore, when it is applied in a model-driven project we expect that the development time is reduced and the optimization of resources used with respect to non-use of the framework.
When our framework can be applied, we expect that our framework allows answering, among others, the following questions: { Does the model describe more information than it is really needed? { Is it possible to evidence whether the identi ed models are for declarative or mapping purposes? { Are the language and notation according to the MDE? { Do the models allow to perform traceability? { Is there any aspect of the IS not covered by the identi ed models? { Can the models generate fully functional software? { Does the model cover a speci c view of the IS? 5
The quality term in MDE context is reduced to two representative trends: one
trend is about the cognitive and semiotic evaluation of notations, usability of
modelling languages and modelling process. As a consecuence, this trend has
several frameworks and guidelines that focus on recomendations for building
models. It is evident frameworks such as SEQUAL[
On the other hand, the quality models term is used to justify metamodels
that relate to the conceptual set of software quality assurance, according with
ISO 9126 or ISO 25000 standards. In [
Most of the above works do not cover the quality of languages and models
from a MDE viewpoint, i.e., they do not explain how multiple proposals for
managing multiple views in a MDE scenario can co-exist. In [
The author thanks Prs Oscar Pastor and Sergio Espan~a (PROS Research Centre, Universitat Politecnica de Valencia, Spain) for their collaborations and their valuable feedback. The author thanks to Colciencias (Colombia) for funding of this work through Colciencias Grant call 512-2010, and Spanish Ministry of Science and Innovation project PROS-Req (TIN2010-19130-C02-02).