bjoernp/tagesschau-2018-2023

---
dataset_info:
features:
- name: date
dtype: string
- name: headline
dtype: string
- name: short_headline
dtype: string
- name: short_text
dtype: string
- name: article
dtype: string
- name: link
dtype: string
splits:
- name: train
num_bytes: 107545823
num_examples: 21847
download_size: 63956047
dataset_size: 107545823
language:
- de
size_categories:
- 10K<n<100K
---

# Tagesschau Archive Article Dataset

A scrape of Tagesschau.de articles from 01.01.2018 to 26.04.2023. Find all source code in [github.com/bjoernpl/tagesschau](https://github.com/bjoernpl/tagesschau).

## Dataset Information
CSV structure:

| Field | Description |
| --- | --- |
| `date` | Date of the article |
| `headline` | Title of the article |
| `short_headline` | A short headline / Context |
| `short_text` | A brief summary of the article |
| `article` | The full text of the article |
| `href` | The href of the article on tagesschau.de |

Size:

The final dataset (2018-today) contains 225202 articles from 1942 days. Of these articles only
21848 are unique (Tagesschau often keeps articles in circulation for ~1 month). The total download
size is ~65MB.

Cleaning:

- Duplicate articles are removed
- Articles with empty text are removed
- Articles with empty short_texts are removed
- Articles, headlines and short_headlines are stripped of leading and trailing whitespace

More details in [`clean.py`](https://github.com/bjoernpl/tagesschau/blob/main/clean.py).

原始信息汇总

Tagesschau Archive Article Dataset 概述

数据集特征

• date：文章日期，数据类型为字符串。
• headline：文章标题，数据类型为字符串。
• short_headline：简短标题/上下文，数据类型为字符串。
• short_text：文章简短摘要，数据类型为字符串。
• article：文章全文，数据类型为字符串。
• link：文章链接，数据类型为字符串。

数据集分割

• train：训练集，包含21,847个样本，总大小为107,545,823字节。

数据集大小

• 下载大小：63,956,047字节。
• 数据集总大小：107,545,823字节。

语言

• de：德语。

数据集规模

• 10K<n<100K：数据集规模介于10,000到100,000之间。