AutoIntent/dstc3

这是一个文本分类数据集，旨在用于机器学习和研究实验。该数据集是通过格式化另一个公开数据集以兼容AutoIntent库而获得的。

This is a text classification dataset intended for machine learning research and experimentation. The dataset is obtained by formatting another publicly available dataset to be compatible with the AutoIntent Library. The dataset includes two configurations: default and intents. The default configuration contains two features: utterance and label, divided into train and oos splits. The intents configuration includes six features: id, name, tags, regexp_full_match, regexp_partial_match, and description, with only one intents split. The dataset is sourced from marcel-gohsen/dstc3 and formatted using the AutoIntent Library.

搜集汇总

背景与挑战

背景概述

该数据集是一个多标签文本分类数据集，用于意图识别任务，包含约15,000条英文utterance，每条utterance对应一个长度为7的二进制标签序列。数据集来源于公开的dstc3数据，经过AutoIntent库格式化处理，分为训练集和测试集，适用于机器学习研究和实验。