The dataset viewer is not available for this split.
Error code: TooBigContentError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
AMNA News Dataset
Dataset Info
This dataset consists of a large-scale collection of Greek news articles scraped from the official website of the Athens-Macedonian News Agency (AMNA: https://www.amna.gr). It includes articles covering a wide range of topics such as politics, economy, science, culture, and international news. Each entry provides the full article text along with its corresponding publication timestamp, category, and original source URL. The dataset spans from June 29, 2011 to March 30, 2026.
Metadata Info
| Column | Type | Nullable | Description |
|---|---|---|---|
id |
string | No | Unique article identifier from the AMNA website |
category |
string | No | Topic category (Πολιτική, Κόσμος, Ελλάδα, Οικονομία, Πολιτισμός, Επιστήμη) |
title |
string | No | The headline or title of the article |
abstract |
string | No | Short summary/lead of the article (empty for ~70% of entries) |
text_clean |
string | No | The full body content of the article in plain text |
author |
string | No | Author name (empty for ~85% of entries) |
created |
datetime | No | Original publication date and time |
modified |
datetime | No | Last modification date and time |
tags |
string | No | Comma-separated tags (empty for ~90% of entries) |
url |
string | No | The direct permanent link to the original article on amna.gr |
text_markdown |
string | No | The full body content of the article in markdown format |
Dataset Statistics:
- Entries: 341,762
- Tokens (BERT): 158,226,440
- Dataset Size: 1.46 GB
- Categories: 6 (Politics: 100,035 | World: 99,973 | Greece: 74,889 | Economy: 41,032 | Culture: 17,624 | Science: 8,212)
- Date Range: June 29, 2011 -- March 30, 2026
Known Data Quality Notes
The following are characteristics of the source data as published on the AMNA website. They have been preserved as-is to maintain reproducibility:
- Test/placeholder articles (14 rows): A small number of articles appear to be test entries published on the source website (e.g., titles such as
336x280,b,e,ααα,ggg). These are not scraping artifacts. - Empty body text (145 rows): Some articles have no body content. These articles exist on the source website with empty or missing text.
- Question marks as quotes (711 rows): Some article texts contain patterns like
?Google+?where question marks appear in place of quotation marks. This is how the content appears on the original website. - Modified before created (89 rows): A small number of entries have a
modifiedtimestamp earlier than theircreatedtimestamp, as recorded by the source CMS. - Author field with URL (1 row): One entry (id
151697) has a URL instead of an author name in theauthorfield. - Sparse optional fields: The
abstract,author, andtagsfields are empty for the majority of entries, reflecting the source data.
Usage
This dataset is useful for:
- Greek NLP tasks (e.g., text classification, summarization, NER)
- Temporal analysis of public communications and news coverage
- Training language models on Greek formal/journalistic text
- Media research on Greek news agency output
License:
This dataset is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0). Any commercial use requires prior written permission from the Athens-Macedonian News Agency (AMNA). All intellectual property rights remain with the Athens-Macedonian News Agency (AMNA).
Contact:
For feedback contact: glossapi.team@eellak.gr
- Downloads last month
- 16