ARCADIA: Scalable Causal Discovery for Corporate Bankruptcy Analysis Using Agentic AI

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Maturo, Fabrizio, Riccio, Donato, Mazzitelli, Andrea, Bifulco, Giuseppe, Paolone, Francesco, Brezeanu, Iulia
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909934839398400
author Maturo, Fabrizio
Riccio, Donato
Mazzitelli, Andrea
Bifulco, Giuseppe
Paolone, Francesco
Brezeanu, Iulia
author_facet Maturo, Fabrizio
Riccio, Donato
Mazzitelli, Andrea
Bifulco, Giuseppe
Paolone, Francesco
Brezeanu, Iulia
contents This paper introduces ARCADIA, an agentic AI framework for causal discovery that integrates large-language-model reasoning with statistical diagnostics to construct valid, temporally coherent causal structures. Unlike traditional algorithms, ARCADIA iteratively refines candidate DAGs through constraint-guided prompting and causal-validity feedback, leading to stable and interpretable models for real-world high-stakes domains. Experiments on corporate bankruptcy data show that ARCADIA produces more reliable causal graphs than NOTEARS, GOLEM, and DirectLiNGAM while offering a fully explainable, intervention-ready pipeline. The framework advances AI by demonstrating how agentic LLMs can participate in autonomous scientific modeling and structured causal inference.
format Preprint
id arxiv_https___arxiv_org_abs_2512_00839
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ARCADIA: Scalable Causal Discovery for Corporate Bankruptcy Analysis Using Agentic AI
Maturo, Fabrizio
Riccio, Donato
Mazzitelli, Andrea
Bifulco, Giuseppe
Paolone, Francesco
Brezeanu, Iulia
Artificial Intelligence
Computation
Methodology
62A99, 68T99, 68T05
I.2.6; I.2.0; I.2.4; G.3
This paper introduces ARCADIA, an agentic AI framework for causal discovery that integrates large-language-model reasoning with statistical diagnostics to construct valid, temporally coherent causal structures. Unlike traditional algorithms, ARCADIA iteratively refines candidate DAGs through constraint-guided prompting and causal-validity feedback, leading to stable and interpretable models for real-world high-stakes domains. Experiments on corporate bankruptcy data show that ARCADIA produces more reliable causal graphs than NOTEARS, GOLEM, and DirectLiNGAM while offering a fully explainable, intervention-ready pipeline. The framework advances AI by demonstrating how agentic LLMs can participate in autonomous scientific modeling and structured causal inference.
title ARCADIA: Scalable Causal Discovery for Corporate Bankruptcy Analysis Using Agentic AI
topic Artificial Intelligence
Computation
Methodology
62A99, 68T99, 68T05
I.2.6; I.2.0; I.2.4; G.3
url https://arxiv.org/abs/2512.00839