From Rule-Based NLP to Prompt-Guided Multi-LLM Pipelines for Text-to-UML and Code Generation
Engineering, Technology & Applied Science Research · 06.06.2026 · 16 (3) · 35287-35294
DOI: 10.48084/etasr.18049 ↗Kurz gefasst
Fachartikel in der Zeitschrift Engineering, Technology & Applied Science Research, online erschienen am 06.06.2026. Verfasst von Zakaria Babaalla und 3 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Automatic transformation of textual specifications into formal software models is a key challenge in model-driven engineering. Despite progress, existing approaches remain fragmented and rarely integrate into a coherent methodological framework. This article offers a progressive and analytical review of UML and text-based code generation methods, structured around three successive contributions reflecting the natural evolution of the field: a first approach based on linguistic analysis and rules, a second based on language models trained on an annotated corpus, and a third relying on prompt-guided LLMs within a multi-model MDA pipeline stabilized by an intermediate Domain-Specific Language (DSL). A comparative study underscores the respective advantages and shortcomings of the considered paradigms and demonstrates that DSL-guided multi-LLM orchestration markedly enhances structural consistency, multilingual robustness, and the quality of generated code.
Zitieren
APA
Babaalla, Z., Abdelmalek, H., Jakimi, A., & Saadane, R. (2026). From Rule-Based NLP to Prompt-Guided Multi-LLM Pipelines for Text-to-UML and Code Generation. Engineering, Technology & Applied Science Research, 16(3), 35287-35294. https://doi.org/10.48084/etasr.18049
Vancouver
Babaalla Z, Abdelmalek H, Jakimi A, Saadane R. From Rule-Based NLP to Prompt-Guided Multi-LLM Pipelines for Text-to-UML and Code Generation. Engineering, Technology & Applied Science Research. 2026;16(3):35287-35294. doi:10.48084/etasr.18049
BibTeX
@article{Babaalla2026,
author = {Babaalla, Zakaria and Abdelmalek, Hamza and Jakimi, Abdeslam and Saadane, Rachid},
title = {From Rule-Based NLP to Prompt-Guided Multi-LLM Pipelines for Text-to-UML and Code Generation},
journal = {Engineering, Technology & Applied Science Research},
year = {2026},
volume = {16},
number = {3},
pages = {35287--35294},
doi = {10.48084/etasr.18049}
}
EUVIMED