THE USE OF LLMs TO OPTIMIZE THE DEVELOPMENT OF WEB APPLICATIONS BUILT WITH DJANGO: DEVELOPING A COMPARATIVE ANALYSIS BETWEEN LLMs IN BUG IDENTIFICATION AND RESPONSE QUALITY
DOI:
https://doi.org/10.56238/bocav25n78-032Keywords:
LLMS’S, Programming, Django, PythonAbstract
The development of web applications using the Django framework, despite its robust structure and high performance, still faces challenges related to bug identification and code maintenance. With the advancement of artificial intelligence, Large Language Models (LLMs) have emerged as a promising alternative for optimizing tasks such as code generation, review, and bug fixing. This article aims to evaluate how LLMs can assist in software development using the Django Python framework. To achieve this, a quantitative research study was conducted in which two LLMs were subjected to problems related to bugs and improvements in a prototype product registration system developed in Django. The variables analyzed were response time, number of errors in the response, response quality (rated from 1 to 5), clarity of explanation, and justification for the response decision. The results were organized into a comparative table, enabling the analysis of the performance, efficiency, and quality of the responses generated by each model. The research was conducted in the Visual Studio Code environment using Python and Django, while respecting software licenses and scientific integrity. The obtained data indicate significant differences between the tested models, contributing to a more careful selection of LLMs in the Django software development cycle.
References
ZHOU, Xiyu; LIANG, Peng; ZHANG, Beiqi; et al. Exploring the problems, their causes and solutions of AI pair programming: A study on GitHub and Stack Overflow. Journal of Systems and Software, v. 219, n. 112204, 2025. Disponível em: https://jyx.jyu.fi/handle/123456789/99112. DOI: https://doi.org/10.1016/j.jss.2024.112204
XU, Chuyang; et al. FlexFL: Flexible and effective fault localization with open-source large language models. ArXiv:2411.10714, 2025. Disponível em: https://arxiv.org/abs/2411.10714.
LI, Zhenhao; et al. Empirical Evaluation of Large Language Models for Novice Program Fault Localization. In: IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER), 2024. Disponível em: https://ieeexplore.ieee.org/document/10684637.
LIMA, F. R.; AZEVEDO, A. I. R.; BARBOSA, H. F. A influência do GitHub Copilot no ensino e aprendizado de programação: percepções e desafios. In: Anais do XXXIV Simpósio Brasileiro de Informática na Educação (SBIE), 2023, Passo Fundo. p. 123-134. Disponível em: https://sol.sbc.org.br/index.php/sbie/article/view/24567.
CAMPOS, A. G. M.; BARBOSA, M. S. S. Uso de modelos de inteligência artificial na geração de código computacional: um estudo sobre percepção docente. Brazilian Journal of Development, Curitiba, v. 9, n. 10, p. 27997-28013, 2023. DOI: 10.34117/bjdv9n10-007. Disponível em: https://ojs.brazilianjournals.com.br/ojs/index.php/BRJD/article/view/63640.
HUANG, K. et al. A Survey on Evaluation of Large Language Models. arXiv:2307.03109, 2023. Disponível em: https://arxiv.org/abs/2307.03109.
LI, X. et al. Large Language Models for Artificial General Intelligence (AGI): A Survey of Foundational Principles and Approaches. arXiv:2501.03151, 2025. Disponível em: https://arxiv.org/abs/2501.03151.
ZHANG, S. et al. A Survey of GPT-3 Family Large Language Models Including ChatGPT and GPT-4. arXiv:2310.12321, 2023. Disponível em: https://arxiv.org/abs/2310.12321.
AHN, M. et al. Do As I Can, Not As I Say: Grounding Language in Robotic Affordances. arXiv:2204.01691, 2022. Disponível em: https://arxiv.org/abs/2204.01691.
GAO, L. et al. PAL: Program-Aided Language Models. arXiv:2211.10435, 2022. Disponível em: https://arxiv.org/abs/2211.10435.
SCHICK, T. et al. Toolformer: Language Models Can Teach Themselves to Use Tools. arXiv:2302.04761, 2023. Disponível em: https://arxiv.org/abs/2302.04761.
CHEN, M. et al. Evaluating Large Language Models Trained on Code. arXiv:2107.03374, 2021. Disponível em: https://arxiv.org/abs/2107.03374.
FENG, Z. et al. CodeBERT: A Pre-Trained Model for Programming and Natural Languages. arXiv:2002.08155, 2020. Disponível em: https://arxiv.org/abs/2002.08155.
LU, S. et al. CodeXGLUE: A Machine Learning Benchmark Dataset for Code Understanding and Generation. arXiv:2102.04664, 2021. Disponível em: https://arxiv.org/abs/2102.04664.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Cristian Abreu Berredo, Gilbert Correia Fernandes, Dadilton Bastos Melo
This work is licensed under a Creative Commons Attribution 4.0 International License.
Copyright (c). Conjuncture Bulletin (BOCA)
This work is licensed under a Creative Commons Attribution 4.0 International License.
