Dzeniks/hover
收藏资源简介:
--- license: mit task_categories: - text-classification --- # Hover Dataset The Hover dataset is a collection of labeled examples for many-hop fact extraction and claim verification tasks. It contains claims, with each claim labeled as either "Supports" or "Refutes". The dataset was created by Yichen Jiang, Shikha Bordia, Zheng Zhong, Charles Dognin, Maneesh Singh, and Mohit Bansal, and was presented in their paper "HoVer: A Dataset for Many-Hop Fact Extraction and Claim Verification" at the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) [Hover page](https://hover-nlp.github.io/). ## Format The Hover dataset is formatted as a TSV file, with each line containing the following fields: - **Claim:** The text of the claim to be verified. - **Label:** The label for the claim, either "0" for "Supports" or "1" for "Refutes". - **Explanation:** A sentence or phrase explaining why the claim is labeled as such. - **Evidence:** Evidence supporting or refuting the claim, if available. This may be a URL or a short text snippet.
许可证:MIT协议 任务类别: - 文本分类 ## Hover 数据集(Hover Dataset) Hover数据集是面向多跳事实抽取与主张验证任务的标注样本集,收录多条待验证主张,每条主张均被标注为“支持(Supports)”或“驳斥(Refutes)”两类之一。该数据集由Yichen Jiang、Shikha Bordia、Zheng Zhong、Charles Dognin、Maneesh Singh与Mohit Bansal构建,并发表于2020年自然语言处理经验方法会议(EMNLP)的论文《HoVer: A Dataset for Many-Hop Fact Extraction and Claim Verification》,其官方项目主页为[Hover页面](https://hover-nlp.github.io/)。 ## 数据格式 Hover数据集采用TSV文件格式存储,每行包含以下四个字段: - **主张(Claim):** 待验证主张的完整文本内容。 - **标签(Label):** 对应主张的标注标签,其中“0”代表“支持(Supports)”,“1”代表“驳斥(Refutes)”。 - **解释(Explanation):** 用于说明该主张标注理由的语句或短语。 - **证据(Evidence):** 用于支持或驳斥该主张的佐证材料(若有),可为URL链接或简短文本片段。
Hover 数据集概述
数据集描述
- 名称: Hover 数据集
- 目的: 用于多跳事实提取和声明验证任务
- 内容: 包含声明及其标签,标签分为“支持”(0)和“反驳”(1)
- 创建者: Yichen Jiang, Shikha Bordia, Zheng Zhong, Charles Dognin, Maneesh Singh, Mohit Bansal
- 发布: 在2020年自然语言处理实证方法会议(EMNLP)上发布
数据格式
- 文件类型: TSV文件
- 字段:
- Claim: 待验证的声明文本
- Label: 声明的标签,0表示“支持”,1表示“反驳”
- Explanation: 解释声明为何被标记为此标签的句子或短语
- Evidence: 支持或反驳声明的证据,可能为URL或文本片段




