Are diagnostic concepts within the reach of LLMs?
Publication date
2025-11-10
Document type
Konferenzbeitrag
Author
Anna Sztyber-Betley
Chanthery, Elodie
Travé-Massuyès, Louise
Merkelbach, Silke
Kukla, Karol
Glotin, Maxence
Organisational unit
Conference
36th International Conference on Principles of Diagnosis and Resilient Systems (DX 2025) ; Nashville, TN ; September 22–24, 2025
Publisher
Schloss Dagstuhl – Leibniz-Zentrum für Informatik
Series or journal
Open Access Series in Informatics (OASIcs)
Periodical volume
136
Book title
36th International Conference on Principles of Diagnosis and Resilient Systems (DX 2025)
First page
2:1
Last page
2:20
Part of the university bibliography
✅
Language
English
Keyword
Fault diagnosis
Large Language Models (LLMs)
Model Based Diagnosis MSO
Redundancy Relations
Conflicts
Diagnoses
Abstract
Model-based diagnosis is a cornerstone of system health monitoring, allowing for the identification of faulty components based on observed behavior and a formal system model. However, obtaining a useful and reliable model is often an expensive and manual task. While the generation of a formal model was the aim of previous work, in this paper, we propose a methodology to use large language models to generate Minimal Structurally Overdetermined sets (MSOs). MSOs are specific subsets of the model equations from which diagnosis tests can be obtained. We investigate two different directions: (i) the large-language-models' ability to generate MSO sets for hybrid systems, similar to those generated by the well-known Fault Diagnosis Toolbox (FDT) (ii) the automated generation of MSOs for Boolean circuits, as well as the computation of the diagnoses. We thus show how both dynamic and static systems can be analysed by large-language models and how their output can be used for effective fault diagnosis. We evaluate our approach on a set of arithmetic and logic circuits, using OpenAI’s LLMs 4o-mini, o1, and o3-mini.
Description
Open Access publication under the terms of the Creative Commons Attribution 4.0 International license (https://creativecommons.org/licenses/by/4.0/legalcode)
Cite as
Anna Sztyber-Betley, Elodie Chanthery, Louise Travé-Massuyès, Silke Merkelbach, Karol Kukla, Maxence Glotin, Alexander Diedrich, and Oliver Niggemann. Are Diagnostic Concepts Within the Reach of LLMs?. In 36th International Conference on Principles of Diagnosis and Resilient Systems (DX 2025). Open Access Series in Informatics (OASIcs), Volume 136, pp. 2:1-2:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2025) https://doi.org/10.4230/OASIcs.DX.2025.2
Version
Published version
Access right on openHSU
Metadata only access