EMBO/soda-nel
收藏Hugging Face2025-10-19 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/EMBO/soda-nel
下载链接
链接失效反馈官方服务：
资源简介：
SODA-SPROUT是一个针对命名实体链接任务进行了角色筛选的生物实体数据集。该数据集专门筛选了在科学实验中实际测量或操作的核心生物实体，即具有assayed（测量）和intervention（干预）角色的蛋白质和基因。数据集经过改进，提高了质量，减少了噪声，并专注于对链接任务最重要的生物实体。数据集包含16,532个图表级别的记录，来自3,101篇论文，总共包含184,413个实体。
SODA-SPROUT is a role-filtered named entity linking dataset focusing on biological entities with assayed and intervention roles, representing the core entities actually measured or manipulated in scientific experiments. The dataset has been improved for higher quality, reduced noise, and a focus on the most important biological entities for linking tasks. It contains 16,532 figure-level records from 3,101 papers, totaling 184,413 entities.
提供机构：
EMBO
