THUCNews

从THUCNews中抽取了20万条新闻标题，文本长度在20到30之间。一共10个类别，每类2万条。

以字为单位输入模型，使用了预训练词向量：搜狗新闻 Word+Character 300d。

类别：财经、房产、股票、教育、科技、社会、时政、体育、游戏、娱乐

A total of 200,000 news headlines were extracted from THUCNews, with each headline's length ranging from 20 to 30 characters. The dataset encompasses 10 distinct categories, with exactly 20,000 samples for each category.

Character-level inputs are employed for the model, and pre-trained embeddings: Sougou News Word+Character 300d are utilized.

The 10 categories are listed as: Finance, Real Estate, Stocks, Education, Science and Technology, Society, Political Affairs, Sports, Games, Entertainment.

搜集汇总

背景与挑战

背景概述

THUCNews数据集是一个中文新闻标题分类数据集，包含20万条新闻标题，文本长度统一在20到30之间，分为财经、房产等10个类别，每类2万条，类别分布均衡。该数据集以字为单位处理，并使用了搜狗新闻预训练词向量，适合用于中文自然语言处理任务，如文本分类模型训练。