MetaQA

该数据集包含了电影领域的超过40万个单跳和多跳（最多3跳）问题，分为三个子集：MetaQA-1hop、MetaQA-2hop和MetaQA-3hop。数据集被划分为单跳问题与多跳问题两部分。规模上，该数据集包含超过40万个问题，其任务定位于多跳知识库问答。

This dataset contains over 400,000 single-hop and multi-hop (up to 3 hops) questions in the film domain, and is divided into three subsets: MetaQA-1hop, MetaQA-2hop and MetaQA-3hop. The dataset is further categorized into two groups: single-hop questions and multi-hop questions. With a total of more than 400,000 questions, this dataset is targeted at the task of multi-hop knowledge base question answering.

搜集汇总

背景与挑战

背景概述

MetaQA是一个电影文本和音频问答基准数据集，包含纯文本、神经翻译模型生成的文本和音频数据三个主要组件，分为1-hop、2-hop和3-hop三个部分，并提供知识库和问题类型文件。