NYT-H
收藏arXiv2020-10-30 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2010.16275v1
下载链接
链接失效反馈官方服务：
资源简介：
NYT-H是由苏州大学人工智能研究院创建的一个用于远监督关系抽取的数据集。该数据集基于远监督生成的训练数据，并由人工标注者对测试数据进行标注，以提高评估的准确性和一致性。数据集包含10,065条人工标注的句子，主要用于解决远监督关系抽取中评估不准确的问题。NYT-H的应用领域主要集中在提高远监督关系抽取的评估质量，确保研究结果的可靠性和客观性。
The NYT-H dataset is constructed for distant supervision relation extraction, and was developed by the Institute of Artificial Intelligence at Soochow University. Its training data is generated through distant supervision, while the test data is manually annotated by human annotators to improve the accuracy and consistency of evaluation. The dataset contains 10,065 manually annotated sentences, and is primarily designed to address the issue of inaccurate evaluation in distant supervision relation extraction. The main application scenarios of NYT-H focus on enhancing the evaluation quality of distant supervision relation extraction, ensuring the reliability and objectivity of research findings.
提供机构：
苏州大学人工智能研究院
创建时间：
2020-10-30
