arXiv：cs.CL· Fahmid Shahriar Iqbal, Ritam Dutt, Soumitra Das, Arnav Verma, Sagnik Ray Choudhury·· 4 小时前AI 评分33
大语言模型在时间抽取任务中的多维泛化能力评估
Evaluating Multi-Dimensional Generalization of Large Language Models in Temporal Extraction Tasks
AI 导读
研究评估了多个大语言模型在时间与事件表达抽取任务中的泛化能力，覆盖领域迁移、对抗扰动、组合性和长度增加四个维度。强基础任务表现通常预示更好的泛化，但在显著分布偏移下这一关系减弱；归纳式提示词在各维度表现最稳定，而规模、架构及演绎、溯因提示策略的收益不均且因维度而异。该研究被 AACL-IJCNLP 2026 Findings 接收。
正文
Abstract:Time and event expression extraction are fundamental temporal reasoning tasks, but the problem remains difficult due to annotation ambiguity, domain sensitivity, and unstable model behavior. Existing evaluations focus on in-domain performance, offering limited insight into reliability under distribution shifts. We evaluate multiple model configurations across families, architectures, and reasoning strategies over four dimensions of generalization, examining transfer from base performance, cross-dimensional correlations, and the effects of scale, architecture, and prompting. This provides a systematic study of how prompted LLMs generalize in time and event expression extraction tasks. We find that strong base-task performance generally predicts better generalization. However, this relationship weakens under substantial distribution shifts. Inductive prompting performs most consistently across domain shift, adversarial perturbations, compositionality, and length increase, while gains from scale, architecture, and deductive and abductive prompting strategies are uneven and dimension-specific. We conclude that LLM generalization in temporal extraction tasks cannot be predicted from any single dimension alone and cannot be reliably inferred from in-domain or single-dimension evaluations, highlighting the need for reasoning strategies that generalize across dimensions.
| Comments: | accepted AACL-IJCNLP 2026 Findings |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.02549 [cs.CL] |
| (or arXiv:2610.02549v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.02549 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Fahmid Shahriar Iqbal [view email]
[v1]
Thu, 1 Oct 2026 22:35:22 UTC (1,728 KB)
来源：arXiv：cs.CL · arxiv.org