Harnessing Large Language Models for Software Issue Learning and Knowledge Retention.

Saved in:

Bibliographic Details
Title:	Harnessing Large Language Models for Software Issue Learning and Knowledge Retention.
Authors:	CHEN, SZU-YU¹ adsylviata@gmail.com, HUANG, NAN-JUNG¹ njhuang.cs11@nycu.edu.tw, KOONG, CHORNG-SHIUH² csko@mail.ntcu.edu.tw, HUANG, SHIH-KUN¹ skhuang@nycu.edu.tw
Source:	Journal of Information Science & Engineering. May2026, Vol. 42 Issue 3, p703-726. 24p.
Subjects:	Language models, Defect tracking (Computer software development), Intelligent tutoring systems, Memory, Software engineering, Computer programming education, Generative adversarial networks, Instructional systems
Abstract:	Project issues often capture the challenges and solutions encountered during software development, making them valuable learning resources for avoiding redundant problem-solving efforts. However, the volume and complexity of issue data present a significant challenge for effective learning. This study proposes an AI-assisted approach to issue-based learning. Leveraging a large language model (LLM) for issue classification and filtering and drawing inspiration from the generative component of Generative Adversarial Networks (GANs) (while excluding the adversarial aspect), we generate plausible yet incorrect answer options to enhance question difficulty. This design supports a gradual, layered learning process. Using open-source project issues from GitHub as the data source, we developed a learning platform called SPILSG to generate and deliver learning materials. Experimental results demonstrate the following: (1) The GPT-4-0613 model achieved the highest compliance rates at 37.4% and 50.6%; (2) In terms of reading effectiveness, the LLM-based SP1LSG system significantly improved novices' understanding of software project issues; and (3) Usability evaluations indicated a generally positive user experience. In summary, AI-assisted learning offers a time-efficient way for learners to comprehend project contexts, familiarize themselves with codebases, and enhance their programming proficiency. [ABSTRACT FROM AUTHOR]
	Copyright of Journal of Information Science & Engineering is the property of Institute of Information Science, Academia Sinica and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Project issues often capture the challenges and solutions encountered during software development, making them valuable learning resources for avoiding redundant problem-solving efforts. However, the volume and complexity of issue data present a significant challenge for effective learning. This study proposes an AI-assisted approach to issue-based learning. Leveraging a large language model (LLM) for issue classification and filtering and drawing inspiration from the generative component of Generative Adversarial Networks (GANs) (while excluding the adversarial aspect), we generate plausible yet incorrect answer options to enhance question difficulty. This design supports a gradual, layered learning process. Using open-source project issues from GitHub as the data source, we developed a learning platform called SPILSG to generate and deliver learning materials. Experimental results demonstrate the following: (1) The GPT-4-0613 model achieved the highest compliance rates at 37.4% and 50.6%; (2) In terms of reading effectiveness, the LLM-based SP1LSG system significantly improved novices' understanding of software project issues; and (3) Usability evaluations indicated a generally positive user experience. In summary, AI-assisted learning offers a time-efficient way for learners to comprehend project contexts, familiarize themselves with codebases, and enhance their programming proficiency. [ABSTRACT FROM AUTHOR]
ISSN:	10162364
DOI:	10.6688/JISE.202605_42(3).0013