WiC-TSV
收藏arXiv2021-01-28 更新2024-06-21 收录
下载链接：
https://competitions.codalab.org/competitions/23683
下载链接
链接失效反馈官方服务：
资源简介：
WiC-TSV是由语义网络公司创建的多领域评估基准，用于词义消歧。该数据集包含3832条记录，覆盖了鸡尾酒、医学和计算机科学三个特定领域。数据集的创建过程涉及从多个资源中提取和验证数据，确保了数据的质量和多样性。WiC-TSV旨在评估模型在无外部词义库存的情况下，对文本中词语的词义进行验证的能力，特别适用于需要高度灵活性和跨领域应用的模型评估。
WiC-TSV is a multi-domain evaluation benchmark created by Semantic Network Corporation for word sense disambiguation. This dataset contains 3,832 records, covering three specific domains: cocktails, medicine, and computer science. The dataset's creation process involves extracting and validating data from multiple sources, ensuring the quality and diversity of the dataset. WiC-TSV aims to evaluate a model's ability to verify the word sense of words in text without relying on external word sense inventories, and is particularly suitable for evaluating models that require high flexibility and cross-domain applicability.
提供机构：
语义网络公司
创建时间：
2020-05-01
