MEDIQA-CORR 2024 Dataset

该数据集整合了来自MS和UW收藏的临床文本，用于评估一个两阶段模型在临床段落文本中进行单字错误纠正的效果。数据集包含了错误标志、错误句子索引以及相应的正确句子，这些信息用于评估和模型训练。在规模上，训练集包含了2,189篇文本；验证集分别从MS和UW中提取了574篇和160篇文本；测试集则包含了925篇文本。任务包括：1) 错误标志预测，2) 错误句子索引检测，以及3) 生成正确句子。

This dataset integrates clinical texts collected from MS and UW, and is used to evaluate the performance of a two-stage model for single-character error correction in clinical paragraph texts. The dataset includes error flags, erroneous sentence indices, and corresponding correct sentences, which are utilized for model training and evaluation. In terms of scale, the training set contains 2,189 texts; the validation set consists of 574 texts from MS and 160 texts from UW respectively; the test set includes 925 texts. The tasks are as follows: 1) Error flag prediction, 2) Erroneous sentence index detection, and 3) Generation of correct sentences.

搜集汇总

背景与挑战

背景概述

MEDIQA-CORR 2024 Dataset是一个专注于临床笔记医学错误识别与纠正的数据集，包含来自MS和UW的多个子集，任务涵盖错误标记、错误句子定位和纠正生成，评估采用准确率和多种NLG指标。