ADAPTIVE SELF-HEALING SOFTWARE SYSTEMS USING LARGE LANGUAGE MODELS: BRIDGING CONFIGURATION VALIDATION, VULNERABILITY DETECTION, AND SPECIFICATION LEARNING
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY · 19.11.2024 · 15 (5) · 1197-1210
DOI: 10.34218/ijcet_15_05_112 ↗Kurz gefasst
Fachartikel in der Zeitschrift INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY, online erschienen am 19.11.2024. Verfasst von Somasekhar Gubbala. Die Arbeit wurde bisher 5-mal zitiert (Crossref).
Abstract (Originalsprache)
Software reliability is a persistent challenge due to dynamic misconfigurations, hidden vulnerabilities, and evolving system specifications. Although recent advances in Large Language Models (LLMs) have shown great potential in various software engineering tasks—such as configuration validation, vulnerability detection, and specification generation—each of these works tends to focus on isolated aspects of reliability. In this paper, we present AutoHeal, a novel LLM-driven framework for adaptive self-healing software systems that unifies these capabilities. AutoHeal employs multi-phase reasoning: (1) validating configurations using prompt-conditioned analysis, (2) detecting vulnerabilities through contextual code exploration, and (3) reconstructing specifications to maintain system consistency. We introduce an LLM-ensemble learning approach that adaptively tunes prompts and responses based on system feedback and historical corrective actions. Experimental results on ten open-source projects demonstrate that AutoHeal achieves an average precision improvement of 18% over state-of-the-art single-task baselines in identifying and automatically remediating misconfigurations and vulnerabilities. This work opens a new direction in building resilient and intelligent self-managed software systems powered by large-scale language models.
Zitieren
APA
Gubbala, S. (2024). ADAPTIVE SELF-HEALING SOFTWARE SYSTEMS USING LARGE LANGUAGE MODELS: BRIDGING CONFIGURATION VALIDATION, VULNERABILITY DETECTION, AND SPECIFICATION LEARNING. INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY, 15(5), 1197-1210. https://doi.org/10.34218/ijcet_15_05_112
Vancouver
Gubbala S. ADAPTIVE SELF-HEALING SOFTWARE SYSTEMS USING LARGE LANGUAGE MODELS: BRIDGING CONFIGURATION VALIDATION, VULNERABILITY DETECTION, AND SPECIFICATION LEARNING. INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY. 2024;15(5):1197-1210. doi:10.34218/ijcet_15_05_112
BibTeX
@article{Gubbala2024,
author = {Gubbala, Somasekhar},
title = {ADAPTIVE SELF-HEALING SOFTWARE SYSTEMS USING LARGE LANGUAGE MODELS: BRIDGING CONFIGURATION VALIDATION, VULNERABILITY DETECTION, AND SPECIFICATION LEARNING},
journal = {INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING AND TECHNOLOGY},
year = {2024},
volume = {15},
number = {5},
pages = {1197--1210},
doi = {10.34218/ijcet_15_05_112}
}
EUVIMED