DSL-Xpert 2.0: Enhancing LLM-driven code generation for domain-specific languages

UDC.coleccionInvestigación
UDC.departamentoCiencias da Computación e Tecnoloxías da Información
UDC.endPage15
UDC.grupoInvLaboratorio de Bases de Datos (LBD)
UDC.institutoCentroCITIC - Centro de Investigación de Tecnoloxías da Información e da Comunicación
UDC.issue107954
UDC.journalTitleInformation and Software Technology
UDC.startPage1
UDC.volume190
dc.contributor.authorLamas Sardiña, Víctor Juan
dc.contributor.authorGarcía-González, Daniel
dc.contributor.authorSala, Luca
dc.contributor.authorRodríguez Luaces, Miguel
dc.date.accessioned2025-11-26T18:35:40Z
dc.date.available2025-11-26T18:35:40Z
dc.date.issued2026-02
dc.descriptionThe tool is available to download at https://github.com/lbdudc/dsl-xpert, along with the user manual. Regarding the text-to-DSL experiments, the framework used is available at https://github.com/lucasala1997/text-to-DSL.
dc.description.abstract[Abstract]: Context: Domain-specific languages (DSLs) are crucial for modeling specialized concepts and offer more fluency and efficiency compared to general-purpose languages. However, their adoption is often slowed by steep learning curves, limited tools, and complex implementations. While large language models (LLMs) have the ability to generate DSL code from natural language, their effectiveness in niche areas is limited due to insufficient training on specific DSL definitions. Objectives: This paper introduces DSL-Xpert 2.0, a tool designed to overcome these challenges by utilizing LLMs to generate DSL code with ease. Methods: By integrating grammar prompting and few-shot learning, the tool effectively handles proprietary DSLs. Additionally, advanced features such as automatic grammar validation, input/output correction, and integration with platforms like OpenAI, HuggingFace, and WebLLM ensure robust and reliable results. These features also simplify workflows for both novice and expert users. The paper further demonstrates the tool's practical value with a running example that illustrates its workflow and includes a complementary user survey conducted across various DSLs of differing complexity. This survey, based on the Technology Acceptance Model (TAM), helps assess the tool’s impact on reducing the DSL learning curve. Results: DSL-Xpert 2.0’s user-friendly and flexible design supports a broad range of DSLs with minimal configuration. Its intuitive interface allows developers to focus on problem-solving, rather than technical issues. User survey results indicate that DSL-Xpert 2.0 significantly reduces the learning effort needed to work with DSLs, and it is perceived as both useful and easy to use. The paper also includes a detailed performance analysis across various LLMs, demonstrating the tool's adaptability and effectiveness. Conclusion: By simplifying DSL development and lowering the barriers to entry, DSL-Xpert 2.0 accelerates the adoption and innovation of DSLs, making it a valuable resource for domain-specific projects.
dc.description.sponsorshipFunding for open access charge: Universidade da Coruña/CISUG. CITIC, as a center accredited for excellence within the Galician University System and a member of the CIGUS Network, has been partially funded by GAIN/Xunta de Galicia: [ED431G 2023/01]; the authors have been partially funded by MCIN/AEI/10.13039/501100011033 and “NextGenerationEU”/PRTR: [PLAGEMIS: TED2021-129245B-C21]; and MCIN/AEI/ 10.13039/501100011033 and EU/ERDF A way of making Europe: [EarthDL: PID2022-141027NB-C21]. GRC: ED431C 2025/34, partially funded by GAIN/Xunta de Galicia.
dc.description.sponsorshipFinanciado para publicación en acceso aberto: Universidade da Coruña/CISUG
dc.description.sponsorshipXunta de Galicia; ED431G 2023/01
dc.description.sponsorshipXunta de Galicia; Xunta de Galicia
dc.description.urihttps://github.com/lbdudc/dsl-xpert
dc.description.urihttps://github.com/lucasala1997/text-to-DSL
dc.identifier.citationLamas, V., Garcia-Gonzalez, D., Sala, L., & Luaces, M. R. (2025). DSL-Xpert 2.0: Enhancing LLM-Driven code generation for domain-specific languages. Information and Software Technology, 107954. https://doi.org/10.1016/j.infsof.2025.107954
dc.identifier.doi10.1016/j.infsof.2025.107954
dc.identifier.issn0950-5849
dc.identifier.issn1873-6025
dc.identifier.urihttps://hdl.handle.net/2183/46557
dc.language.isoeng
dc.publisherElsevier
dc.relation.projectIDinfo:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023/TED2021-129245B-C21/ES/PLATAFORMA PARA LA GENERACIÓN AUTOMÁTICA DE SISTEMAS DE INFORMACIÓN DE LA MOVILIDAD ENERGÉTICAMENTE EFICIENTES, BASADOS EN ESTRUCTURAS DE DATOS COMPACTAS Y GIS (PLAGEMIS)
dc.relation.projectIDinfo:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica, Técnica y de Innovación 2021-2023/PID2022-141027NB-C21/ES/MODELADO, DESCUBRIMIENTO, EXPLORACION Y ANALISIS DE DATA LAKES MEDIOAMBIENTALES
dc.relation.urihttps://doi.org/10.1016/j.infsof.2025.107954
dc.rightsAttribution-NonCommercial-NoDerivatives 4.0 Internationalen
dc.rights.accessRightsopen access
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/4.0/
dc.subjectDomain-specific languages (DSLs)
dc.subjectLarge language models (LLMs)
dc.subjectSemantic parsing
dc.subjectGrammar prompting
dc.subjectFew-shot learning
dc.titleDSL-Xpert 2.0: Enhancing LLM-driven code generation for domain-specific languages
dc.typejournal article
dc.type.hasVersionVoR
dspace.entity.typePublication
relation.isAuthorOfPublicationf5f01d97-f28d-46b7-b99d-3bd795ee2677
relation.isAuthorOfPublication21118aa4-994c-45e0-aff0-4750156048b5
relation.isAuthorOfPublicationfbde3bd9-d786-4ef0-89ec-6af2091fa415
relation.isAuthorOfPublication.latestForDiscoveryf5f01d97-f28d-46b7-b99d-3bd795ee2677

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Lamas_Victor_2025_DSL_Xpert_2_0.pdf
Size:
3.26 MB
Format:
Adobe Portable Document Format