glossAPI/amna-press
收藏资源简介:
--- license: cc-by-nc-nd-4.0 language: - el --- ## 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](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 `modified` timestamp earlier than their `created` timestamp, 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 the `author` field. - **Sparse optional fields:** The `abstract`, `author`, and `tags` fields 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
许可证:CC BY-NC-ND 4.0 语言: - el **AMNA新闻数据集** **数据集概况** 本数据集收录了从雅典-马其顿通讯社(Athens-Macedonian News Agency,简称AMNA)官方网站[https://www.amna.gr](https://www.amna.gr)爬取的大规模希腊语新闻文章集合。数据集涵盖政治、经济、科学、文化、国际新闻等多元主题,每条数据均包含完整文章文本、对应发布时间戳、分类标签及原始来源URL。数据集的时间跨度为2011年6月29日至2026年3月30日。 **元数据信息** | 列名 | 数据类型 | 是否可空 | 描述 | | :--- | :--- | :--- | :--- | | `id` | 字符串 | 否 | AMNA官网唯一文章标识符 | | `category` | 字符串 | 否 | 主题分类(Πολιτική(政治)、Κόσμος(世界新闻)、Ελλάδα(希腊国内)、Οικονομία(经济)、Πολιτισμός(文化)、Επιστήμη(科学)) | | `title` | 字符串 | 否 | 文章标题 | | `abstract` | 字符串 | 否 | 文章简短摘要/导语(约70%的条目为空) | | `text_clean` | 字符串 | 否 | 纯文本格式的完整文章正文 | | `author` | 字符串 | 否 | 作者姓名(约85%的条目为空) | | `created` | 日期时间 | 否 | 原始发布日期与时间 | | `modified` | 日期时间 | 否 | 最后修改日期与时间 | | `tags` | 字符串 | 否 | 逗号分隔的标签(约90%的条目为空) | | `url` | 字符串 | 否 | amna.gr官网原文的永久直接链接 | | `text_markdown` | 字符串 | 否 | Markdown格式的完整文章正文 | **数据集统计信息** - 数据条目数:341,762 - BERT Token数:158,226,440 - 数据集大小:1.46 GB - 分类分布:共6类(政治:100,035 | 世界新闻:99,973 | 希腊国内:74,889 | 经济:41,032 | 文化:17,624 | 科学:8,212) - 时间范围:2011年6月29日 —— 2026年3月30日 **已知数据质量说明** 以下为AMNA官网源数据的固有特征,为保证实验可复现性,我们严格保留源数据原貌未做修改: - **测试/占位文章(14条)**:少量条目为源网站发布的测试内容(例如标题为`336x280`、`b`、`e`、`ααα`、`ggg`的文章),此类内容并非爬取过程产生的异常数据。 - **空正文文本(145条)**:部分文章无正文内容,此类条目在源网站上本身即为文本缺失或空白状态。 - **引号替换为问号(711条)**:部分文章文本存在`?Google+?`这类模式,即用问号替代引号,这与源网站上的原文格式完全一致。 - **修改时间早于发布时间(89条)**:少量条目存在`modified`时间戳早于`created`时间戳的情况,与源内容管理系统(Content Management System,简称CMS)的记录一致。 - **作者字段含URL(1条)**:1条ID为`151697`的条目,其`author`字段内容为URL而非作者姓名。 - **稀疏可选字段**:`abstract`、`author`及`tags`字段在绝大多数条目中均为空,这与源数据的实际情况一致。 **应用场景** 本数据集可应用于以下场景: - 希腊语自然语言处理(Natural Language Processing,简称NLP)任务(例如文本分类、摘要生成、命名实体识别(Named Entity Recognition,简称NER)) - 公共传播与新闻报道的时序分析 - 基于希腊正式/新闻文本的大语言模型(Large Language Model,简称LLM)训练 - 希腊通讯社产出内容的媒体研究 **许可证** 本数据集采用知识共享署名-非商业性使用-禁止演绎4.0国际许可协议(CC BY-NC-ND 4.0)进行授权。任何商业使用均需事先获得雅典-马其顿通讯社(AMNA)的书面许可。所有知识产权归雅典-马其顿通讯社(AMNA)所有。 **联系方式** 如有反馈,请联系:glossapi.team@eellak.gr



