biu-nlp/CoreSearchV2

CoreSearch数据集是一个清理版本，主要用于跨文档事件共指搜索任务。该数据集对事件提及的跨度进行了修正，解决了单个跨度中包含两个事件的情况，并移除了错误的事件提及。数据集支持从大规模文档集合中查找与查询事件共指的所有提及。数据集包含训练集、验证集和测试集，分别包含不同数量的簇和段落。数据集基于Wikipedia内容，并遵循Creative Commons Attribution-ShareAlike 3.0 Unported License。

原始信息汇总

数据集名称

CoreSearch Dataset

数据集版本

A cleaner version of the CoreSearch dataset

数据集清洗内容

• Fix event mention spans
  • 修正事件提及的范围，如将"Academy Awards 65th ceremony"修正为"Academy Awards"。
• Resolve cases of two events in a single span
  • 解决单一范围中包含两个事件的情况，如将"killing or abduction of several hundred villagers"拆分为"killing"。
• Remove erroneous event mentions (Entities)
  • 移除错误的事件提及，如移除"I Want Your Love"（歌曲名）。

数据集语言

英语

数据集加载

可通过Huggingface Hub加载数据集，示例代码如下： python from huggingface_hub import hf_hub_url, cached_download import json REPO_ID = "datasets/Intel/CoreSearchV2" DPR_FILES = "/dpr/"

dpr_files = ["dpr/Dev.json", "dpr/Train.json", "dpr/Test.json"]

dpr_jsons = list() for _file in dpr_files: dpr_jsons.append(json.load(open(cached_download( hf_hub_url(REPO_ID, _file)), "r")))

数据集分割

• Final version of the CD event coreference search dataset
  • Train

    • Clusters: 229

    • Passages (with Mentions): 1,429

    • Added Destructor Passages: 922,736

    • Total Passages: 924,239

  • Valid

    • Clusters: 48

    • Passages (with Mentions): 335

    • Added Destructor Passages: 923,376

    • Total Passages: 923,717

  • Test

    • Clusters: 226

    • Passages (with Mentions): 1,206

    • Added Destructor Passages: 923,746

    • Total Passages: 925,012

  • Total

    • Clusters: 503

    • Passages (with Mentions): 2,970

    • Added Destructor Passages: 2,769,858

    • Total Passages: 2,772,968

引用信息

@inproceedings{eirew-etal-2022-cross, title = "Cross-document Event Coreference Search: Task, Dataset and Modeling", author = "Eirew, Alon and Caciularu, Avi and Dagan, Ido", booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.emnlp-main.58", pages = "900--913", abstract = "The task of Cross-document Coreference Resolution has been traditionally formulated as requiring to identify all coreference links across a given set of documents. We propose an appealing, and often more applicable, complementary set up for the task {--} Cross-document Coreference Search, focusing in this paper on event coreference. Concretely, given a mention in context of an event of interest, considered as a query, the task is to find all coreferring mentions for the query event in a large document collection. To support research on this task, we create a corresponding dataset, which is derived from Wikipedia while leveraging annotations in the available Wikipedia Event Coreferecene dataset (WEC-Eng). Observing that the coreference search setup is largely analogous to the setting of Open Domain Question Answering, we adapt the prominent Deep Passage Retrieval (DPR) model to our setting, as an appealing baseline. Finally, we present a novel model that integrates a powerful coreference scoring scheme into the DPR architecture, yielding improved performance.", }

许可证

数据集提供在<a href="https://creativecommons.org/licenses/by-sa/3.0/deed.en_US">Creative Commons Attribution-ShareAlike 3.0 Unported License</a>下。

搜集汇总

背景与挑战

背景概述

CoreSearchV2是一个用于跨文档事件共指搜索任务的英文数据集，基于Wikipedia构建并经过清理，包含训练、验证和测试分割，用于事件共指检索研究。但数据集存在生成错误，部分数据文件列不匹配，可能影响使用。