trec

TREC数据集，专注于问题分类任务，包含约6000个英文问题样本，分为训练集和测试集。每个问题都带有粗粒度和细粒度的类别标签，分别对应6个和50个不同的类别。数据来源于多个渠道，包括USC发布的英文问题、人工构建的问题以及TREC竞赛的问题，并经过人工标注。该数据集可用于多类别文本分类任务。目前，数据集的授权许可信息未知。

TREC Dataset, which focuses on question classification tasks, contains approximately 6,000 English question samples split into training and test sets. Each question is annotated with coarse-grained and fine-grained category labels, corresponding to 6 and 50 distinct categories respectively. The dataset is sourced from multiple channels, including English questions released by USC, manually constructed questions, and questions from TREC competitions, and has undergone manual annotation. This dataset is applicable to multi-class text classification tasks. Currently, the licensing information of this dataset remains unknown.