Diagnosing Fine-Grained Inconsistency Classification in Financial Disclosure Text
Quick Answer
This study investigates fine-grained inconsistency classification in financial disclosures, achieving 65.3% accuracy with a fine-tuned 300M encoder.
Quick Take
The research highlights the challenges of localization quality, particularly for referential inconsistencies, and suggests that better evidence extraction and reasoning are essential for progress.
Key Points
- Fine-tuned 300M encoder achieved 65.3% accuracy on SBID-FD benchmark.
- Referential inconsistencies are highly sensitive to localization quality.
- Factual and logical inconsistencies remain challenging even with provided evidence.
- Localization errors and type-discrimination errors are distinct issues.
- Improving evidence extraction is crucial for better inconsistency classification.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Financial disclosures contain numerical claims, temporal statements, entity references, policy commitments, and risk descriptions that may conflict in qualitatively different ways. Detecting a conflict is only the first step: review workflows may also need to determine its type, since numerical, temporal, referential, factual, and normative inconsistencies require different evidence and downstream checks. We study this problem as fine-grained inconsistency classification. Using a fixed 5,940-instance snapshot of SBID-FD, a synthetic financial-disclosure benchmark with 11 inconsistency labels and paired reference evidence spans, we compare frozen embedding classifiers, fine-tuned encoders, evidence-augmented classifiers, prompted large language models, and LoRA-adapted generative models under a shared evaluation protocol. A fine-tuned 300M encoder reaches 61.9% accuracy, compared with 61.5% for a LoRA-adapted Qwen3.5-9B model and 61.3% for GPT-5.4. Because these systems differ in architecture, supervision, training objective, and input format, we interpret this as a practical efficiency result for compact supervised encoders rather than a controlled conclusion about model scale. Supplying gold evidence spans improves the fine-tuned encoder to 65.3%, whereas automatically predicted spans recover a meaningful but incomplete share of that gain, indicating that localization quality remains a bottleneck. Class-level analyses show that Referential inconsistencies are especially sensitive to localization quality, while Factual and Logical inconsistencies remain difficult even when the relevant evidence is provided. Together, the oracle, distractor, and per-class analyses separate localization errors from residual type-discrimination errors, indicating that progress requires both stronger evidence extraction and better reasoning over closely related inconsistency categories.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.26368 [cs.CL] |
| (or arXiv:2607.26368v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.26368 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Aman Kumar [view email]
[v1]
Wed, 29 Jul 2026 01:03:53 UTC (42 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.