Empirical assessment of advantages and disadvantages of model transformation languages

Loading...
Thumbnail Image

Date

2023-11-10

Authors

Höppner, Stefan

Journal Title

Journal ISSN

Volume Title

Publication Type

Dissertation

Published in

Abstract

Context: Model-driven software engineering envisages the use of model transformations to evolve models. Model transformation languages, domain-specific languages developed for this task, are touted with many benefits over general-purpose languages. However, most of these claims have neither been substantiated nor investigated thoroughly. Moreover, they frequently lack the contextual information required to critically assess their merit or build meaningful empirical studies around them. Objective: The main objective of this thesis is to aggregate all necessary data to set up proper evaluation and use this data to asses the most prevalent claims about model transformation languages empirically. We aim to provide evidence on whether those claims withstand rigorous empirical scrutiny. We further want to provide a foundation of data upon which more empirical evaluation can be built. Method: To address our objectives, we employ several research methodologies. Initially, we use a structured literature review to determine the state of research and what claims about quality attributes of MTLs are propagated in literature. The SLR results serve as a basis for conducting semi-structured interviews to collect qualitative data on relevant factors and co-founding factors pertaining to the claims discussed. We quantify the identified influences between quality attributes, factors and co-founding variables using structural equation modelling and an online survey. Finally, we use repository mining and design science to collect and prepare artefacts. The artefacts are used in two separate case studies to empirically evaluate several MTL quality attributes based on the previously identified factors. Results: Our results show that many quality attributes believed to be associated with MTLs are in dire need of empirical evidence. To aid in this task, we contribute a quantified structure model describing factors' influence and moderation effects on quality attributes of MTLs. The model aggregates the results of our literature review, interview study and online survey using structural equation modelling. The literature review produced a comprehensive list of quality attributes for which advantages and disadvantages of MTLs are claimed. The interviews resulted in factors contributing to the perception of quality attributes of MTLs and several co-founding factors defining context for the evaluation thereof. Data from the online survey is then used for quantification. Based on the in-depth discussions during our interviews, we further contribute 15 suggestions on actions requiring community-wide effort to improve confidence in the usefulness of model transformation languages. These are further refined based on the quantitative results of the online survey. We also make a direct contribution through the results of our case studies. The first case study provides empirical evidence of how well the ATL transformation language is suited to a category of model transformation problems. The second case study demonstrates the shortcomings of model transformations written in both legacy and modern Java styles. To execute these evaluations, we developed a novel approach for translating ATL transformations into Java code and a classification schema for model transformations written in Java. Conclusion: Our results demonstrate that empirical evaluation of model transformation languages is feasible and necessary. Efforts to provide more empirical substance need to be undergone, and lacklustre language capabilities and tooling need to be improved. The results of this thesis can provide a basis for these further actions.

Description

Faculties

Fakultät für Ingenieurwissenschaften, Informatik und Psychologie

Institutions

Institut fĂĽr Softwaretechnik und Programmiersprachen

Citation

DFG Project uulm

EU Project THU

Other projects THU

License

CC BY 4.0 International

Is version of

Has version

Supplement to

Supplemented by

Has erratum

Erratum to

Part of

DOI external

DOI external

Institutions

Periodical

Degree Program

DFG Project THU

item.page.thu.projectEU

item.page.thu.projectOther

Series

Keywords

Model driven software engineering, Model transformation language, Advantages, Disadvantages, Interview study, Structured literature review, Repository mining, Case study, Modellgetriebene Entwicklung, Modelltransformation, Nachteil, Fallstudie, Case studies, Software engineering, DDC 000 / Computer science, information & general works