jnlpba

JNLPBA数据集专注于生物医学领域的命名实体识别任务，它基于GENIA 3.02语料库，该语料库包含从MEDLINE检索的2000篇摘要，并根据化学分类进行了人工标注。数据集包含英文文本，规模在1万到10万条样本之间，其中训练集包含18546个样本，验证集包含3856个样本。数据集中，每个样本都包含句子ID、分词后的tokens以及对应的命名实体标签，标签采用BIO标注体系，标注了DNA、RNA、细胞系、细胞类型和蛋白质等生物实体。该数据集支持命名实体识别任务，但授权许可信息缺失，需要进一步补充。

The JNLPBA dataset focuses on the named entity recognition (NER) task in the biomedical domain. It is based on the GENIA 3.02 corpus, which contains 2000 abstracts retrieved from MEDLINE and manually annotated according to chemical classification. The dataset consists of English texts, with a total scale ranging from 10,000 to 100,000 samples. Specifically, the training set contains 18,546 samples and the validation set includes 3,856 samples. Each sample in the dataset includes a sentence ID, tokenized tokens, and corresponding named entity tags. The annotation follows the BIO tagging scheme, covering biomedical entities such as DNA, RNA, cell lines, cell types, and proteins. This dataset supports named entity recognition tasks, but its license information is missing and requires further supplementation.

搜集汇总

背景与挑战

背景概述

JNLPBA是一个专注于生物医学领域的命名实体识别数据集，基于GENIA 3.02语料库，包含从MEDLINE检索的2000篇摘要，并人工标注了DNA、RNA、细胞系、细胞类型和蛋白质等生物实体。该数据集采用BIO标注体系，用于训练和验证命名实体识别模型，但授权许可信息目前缺失。