EUVIMEDEUVIMED entdecken
Crossref

Language Models and Logic Programs for Trustworthy Tax Reasoning

William Jurayj, Nils Holzenberger, Benjamin Van Durme

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (45) · 38688-38698

DOI: 10.1609/aaai.v40i45.41212 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von William Jurayj und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

According to the United States Internal Revenue Service, "the average American spends $270 and 13 hours filing their taxes". Even beyond the U.S., tax filing requires complex reasoning, combining application of overlapping rules with numerical calculations. Because errors can incur costly penalties, any automated system must deliver high accuracy and auditability, making modern large language models (LLMs) poorly suited for this task. We propose an approach that integrates LLMs with a symbolic solver to calculate tax obligations. We evaluate variants of this system on the challenging StAtutory Reasoning Assessment (SARA) dataset, and include a novel method for estimating the cost of deploying such a system based on real-world penalties for tax errors. We further show how combining up-front translation of plain-text rules into formal logic programs, combined with intelligently retrieved exemplars for formal case representations, can dramatically improve performance on this task and reduce costs to well below real-world averages. Our results demonstrate the effectiveness of applying semantic parsing methods to statutory reasoning, and show promising economic feasibility of neuro-symbolic architectures for increasing access to reliable tax assistance.

Zitieren

APA

Jurayj, W., Holzenberger, N., & Durme, B. V. (2026). Language Models and Logic Programs for Trustworthy Tax Reasoning. Proceedings of the AAAI Conference on Artificial Intelligence, 40(45), 38688-38698. https://doi.org/10.1609/aaai.v40i45.41212

Vancouver

Jurayj W, Holzenberger N, Durme BV. Language Models and Logic Programs for Trustworthy Tax Reasoning. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(45):38688-38698. doi:10.1609/aaai.v40i45.41212

BibTeX
@article{Jurayj2026,
  author = {Jurayj, William and Holzenberger, Nils and Durme, Benjamin Van},
  title = {Language Models and Logic Programs for Trustworthy Tax Reasoning},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {45},
  pages = {38688--38698},
  doi = {10.1609/aaai.v40i45.41212}
}
Kostenlos starten →