ParaQA

ParaQA 是一个问答 (QA) 数据集，具有多个转述响应，用于知识图 (KG) 上的单轮对话。该数据集是使用半自动框架创建的，该框架使用反向翻译等技术生成对答案的各种释义。现有的 KG 对话式问答数据集（单轮/多轮）专注于问题释义，最多只提供一个答案语言化。但是，ParaQA 包含 5000 个问答对，每个问题至少有 2 个，最多有 8 个独特的释义回答。ParaQA is a question answering (QA) dataset with multiple paraphrased responses tailored for single-turn dialogue over knowledge graphs (KGs). It was developed using a semi-automated framework that leverages techniques such as back-translation to generate diverse paraphrases for answers. Unlike existing KG-based conversational QA datasets (covering both single-turn and multi-turn scenarios), which primarily focus on paraphrasing the input question and provide at most one verbalized answer per query, ParaQA includes 5000 QA pairs, with each question accompanied by at least 2 and up to 8 distinct paraphrased responses.

搜集汇总

背景与挑战

背景概述

ParaQA是一个用于知识图上单轮对话的问答数据集，其特色在于每个问题都配有2到8个独特的转述答案。该数据集包含5000个问答对，通过半自动框架生成，旨在丰富答案的多样性。