MultiTQ

该数据集以ICEWS知识库为基础，专注于时间知识推理。它不仅包含更常见的事件，还包含一些较为冷门的事件，这为大型语言模型（LLMs）带来了挑战。该数据集规模宏大，包含了超过50,000个问题-答案对，旨在应对时间知识问答任务。

This dataset is built upon the ICEWS knowledge base and focuses on temporal knowledge reasoning. It includes not only prevalent events but also some relatively rare ones, which poses challenges for large language models (LLMs). With a large scale, this dataset contains over 50,000 question-answer pairs, and is designed to tackle the temporal knowledge question answering task.

搜集汇总

背景与挑战

背景概述

MultiTQ是一个大规模的时间问答数据集，包含丰富的相关事实和多种时间粒度，用于知识图谱上的多粒度时间问题回答研究。数据集在Hugging Face平台上公开，适用于自然语言处理和时间推理任务。