arXiv cs.CL· Adam Remaki, Xavier Tannier, Christel G\'erardin·· 10 小时前AI 评分22
LongBEL:面向长上下文与文档一致性的生物医学实体链接框架
LongBEL: Long-Context and Document-Consistent Biomedical Entity Linking
AI 导读
LongBEL 是一个文档级生成式生物医学实体链接框架,将全文上下文与历史预测记忆结合,并用交叉验证预测而非金标签训练记忆模块,以缩小训练与推理的差距、抑制级联错误。在英语、法语、西班牙语的五个生物医学基准上,它优于句子级生成基线,概念在文档内频繁复现的数据集上提升最大。融合局部、全局与记忆三种变体的集成版本在所有基准上取得最佳结果,分析显示增益主要来自重复概念的文档级一致性改善。
正文 · 原文
Abstract:Biomedical entity linking maps textual mentions to concepts in structured knowledge bases such as UMLS or SNOMED CT. Most existing systems link each mention independently, using only the mention or its surrounding sentence. This ignores dependencies between mentions in the same document and can lead to inconsistent predictions, especially when the same concept appears under different surface forms. We introduce LongBEL, a document-level generative framework that combines full-document context with a memory of previous predictions. To make this memory robust, LongBEL is trained with cross-validated predictions rather than gold labels, reducing the mismatch between training and inference and limiting cascading errors. Experiments on five biomedical benchmarks across English, French, and Spanish show that LongBEL improves over sentence-level generative baselines, with the largest gains on datasets where concepts frequently recur within documents. An ensemble of local, global, and memory-based variants achieves the best results across all benchmarks. Further analysis shows that the largest gains occur on recurring concepts, suggesting that LongBEL mainly improves document-level consistency rather than isolated mention disambiguation.
| Comments: | 9 pages, 2 figures |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2605.13451 [cs.CL] |
| (or arXiv:2605.13451v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.13451 arXiv-issued DOI via DataCite |
Submission history
From: Adam Remaki [view email]
[v1]
Wed, 13 May 2026 12:47:21 UTC (15,956 KB)
[v2]
Thu, 19 Nov 2026 17:22:16 UTC (2,008 KB)
来源:arXiv cs.CL · arxiv.org