On the Variations of ChatGPT’s Response Quality for Generating Source Code Across Programming Languages

Ángela González de Diego, Franz Wotawa*

*Korrespondierende/r Autor/-in für diese Arbeit

Publikation: Beitrag in Buch/Bericht/KonferenzbandBeitrag in einem KonferenzbandBegutachtung

Abstract

The rise of Large Language Models, particularly the ChatGPT model, has transformed the field of natural language information processing and has led to widespread adoption in a diverse range of applications and across a multitude of industries. In this paper, we focus on assessing the quality of the responses generated by Chat-GPT for the code generation tasks using seven different programming languages. We selected the languages considering diversity in terms of the fields of application, philosophies, and popularity. We carried out an experimental evaluation utilizing different introductory coding examples for each of the programming languages using the pass@k metric for evaluation. The results indicate a correlation between the effectiveness of the model and the popularity of programming languages.

Originalspracheenglisch
TitelTesting Software and Systems - 36th IFIP WG 6.1 International Conference, ICTSS 2024, Proceedings
Redakteure/-innenHéctor D. Menéndez, Gema Bello-Orgaz, Pepita Barnard, John Robert Bautista, Arya Farahi, Santanu Dash, DongGyun Han, Sophie Fortz, Victor Rodriguez-Fernandez
Herausgeber (Verlag)Springer Science and Business Media Deutschland GmbH
Seiten63-78
Seitenumfang16
ISBN (Print)9783031808883
DOIs
PublikationsstatusVeröffentlicht - 25 Jan. 2025
Veranstaltung36th IFIP WG 6.1 International Conference on Testing Software and Systems, ICTSS 2024 - London, Großbritannien / Vereinigtes Königreich
Dauer: 30 Okt. 20241 Nov. 2024

Publikationsreihe

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band15383 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349

Konferenz

Konferenz36th IFIP WG 6.1 International Conference on Testing Software and Systems, ICTSS 2024
Land/GebietGroßbritannien / Vereinigtes Königreich
OrtLondon
Zeitraum30/10/241/11/24

ASJC Scopus subject areas

  • Theoretische Informatik
  • Allgemeine Computerwissenschaft

Fingerprint

Untersuchen Sie die Forschungsthemen von „On the Variations of ChatGPT’s Response Quality for Generating Source Code Across Programming Languages“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren