cd /news/natural-language-processing/longbel-long-context-and-document-co… · home › topics › natural-language-processing › article
[ARTICLE · art-173076] src=machinebrief.com ↗ pub= topic=natural-language-processing verified=true sentiment=↑ positive

LongBEL: Long-Context and Document-Consistent Biomedical Entity Linking

LongBEL, a document-level generative framework for biomedical entity linking introduced in arXiv paper 2605.13451v2, improves on sentence-level generative baselines across five biomedical benchmarks in English, French, and Spanish by combining full-document context with a memory of previous predictions. The framework trains that memory on cross-validated predictions rather than gold labels to reduce the mismatch between training and inference and limit cascading errors. The largest gains occur on datasets where concepts frequently recur within documents, indicating LongBEL mainly improves document-level consistency rather than isolated mention disambiguation, and an ensemble of local, global, and memory-based variants achieves the best results across all benchmarks.

key takeaway

LongBEL's document-level memory trained on cross-validated predictions rather than gold labels beats sentence-level generative baselines on five biomedical entity-linking benchmarks, with the biggest gains on datasets where concepts recur within documents.

by read1 min views2 publishedNov 20, 2026

arXiv:2605.13451v2 Announce Type: replace 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.

── more in #natural-language-processing 4 stories · sorted by recency
── more on @longbel 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/longbel-long-context…] indexed:0 read:1min 2026-11-20 · —