MenatQA

MenatQA是由中国科学院认知与决策智能实验室创建的新型数据集，包含2853个样本，用于评估大型语言模型的时间理解和推理能力。该数据集涵盖了时间范围因子、顺序因子和反事实因子三个时间敏感因子，旨在解决大型语言模型在时间推理上的不足。数据集的创建过程涉及从TimeQA数据集中提取时间相关问题，并通过特定的标注和脚本进行精细化处理。MenatQA的应用领域主要集中在问答系统，特别是需要时间敏感推理的任务，如历史事件的问答等。

MenatQA is a novel dataset developed by the Laboratory of Cognitive and Decision Intelligence, Chinese Academy of Sciences, which contains 2853 samples for evaluating the temporal comprehension and reasoning capabilities of large language models (LLMs). This dataset covers three time-sensitive factors: temporal scope factor, sequential factor, and counterfactual factor, aiming to address the limitations of large language models in temporal reasoning. The construction of MenatQA involves extracting time-related questions from the TimeQA dataset, followed by fine-grained processing via specific annotation and scripts. The main application domains of MenatQA focus on question answering systems, especially tasks requiring time-sensitive reasoning, such as question answering regarding historical events.

搜集汇总

背景与挑战

背景概述

MenatQA是由中国科学院认知与决策智能实验室创建的数据集，包含2853个样本，专注于评估大型语言模型的时间理解和推理能力，涵盖三个时间敏感因子，主要应用于问答系统中的时间敏感推理任务。