IIRC (Incomplete Information Reading Comprehension)

包含来自英语维基百科段落的超过 13,000 个问题，这些问题仅提供部分信息来回答这些问题，缺失的信息出现在一个或多个链接文档中。这些问题是由无法访问任何链接文档的群众工作人员编写的，导致问题与答案出现的上下文几乎没有词汇重叠。

Contains over 13,000 questions sourced from English Wikipedia paragraphs. Each question only provides partial information for answering, whereas the missing information required to resolve them is contained in one or more linked documents. These questions were authored by crowdworkers who had no access to any of the linked documents, leading to almost no lexical overlap between the questions and the contexts where their corresponding answers are located.

搜集汇总

背景与挑战

背景概述

IIRC 是一个基于英语维基百科的阅读理解数据集，包含超过 1.3 万个问题，其特点是问题仅提供部分信息，缺失内容需通过链接文档获取。该数据集由众包人员在不访问链接文档的情况下编写，旨在减少问题与答案间的词汇重叠，发布于 2020 年。