Title:Argumentative Reasoning with Language Models on Non-factorized Case Bases
Abstract:In this paper, we investigate how language models can perform case-based reasoning (CBR) on non-factorized case bases. We introduce a novel framework, argumentative agentic models for case-based reasoning (AAM-CBR), which extends abstract argumentation for case-based reasoning (AA-CBR). Unlike traditional approaches that require factorization of previous cases, AAM-CBR leverages language models to determine case coverage and extract factors based on new cases. This enables factor-based reasoning without exposing or preprocessing previous cases, thus improving both flexibility and pr…
Title:Argumentative Reasoning with Language Models on Non-factorized Case Bases
Abstract:In this paper, we investigate how language models can perform case-based reasoning (CBR) on non-factorized case bases. We introduce a novel framework, argumentative agentic models for case-based reasoning (AAM-CBR), which extends abstract argumentation for case-based reasoning (AA-CBR). Unlike traditional approaches that require factorization of previous cases, AAM-CBR leverages language models to determine case coverage and extract factors based on new cases. This enables factor-based reasoning without exposing or preprocessing previous cases, thus improving both flexibility and privacy. We also present initial experiments to assess AAM-CBR performance by comparing the proposed framework with a baseline that uses a single-prompt approach to incorporate both new and previous cases. The experiments are conducted based on a synthetic credit card application dataset. The result shows that AAM-CBR surpasses the baseline only when the new case contains a richer set of factors. The finding indicates that language models can handle case-based reasoning with a limited number of factors, but face challenges as the number of factors increase. Consequently, integrating symbolic reasoning with language models, as implemented in AAM-CBR, is crucial for effectively handling cases involving many factors.
| Comments: | Presented at NeLaMKRR@KR, 2025 (arXiv:2511.09575) |
| Subjects: | Logic in Computer Science (cs.LO) |
| Report number: | NeLaMKRR/2025/01 |
| Cite as: | arXiv:2512.12656 [cs.LO] |
| (or arXiv:2512.12656v1 [cs.LO] for this version) | |
| https://doi.org/10.48550/arXiv.2512.12656 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ha Thanh Nguyen [view email] [v1] Sun, 14 Dec 2025 12:06:39 UTC (232 KB)