A Python-based metadata specification for exchanging news articles between different systems. The purpose of this is to differentiate between a media crawling/extraction and mining/NLP phase of a media analytics system.
articledata files are line-delimited JSON files with each row describing an object with the structure detailed in article.py. It's a convention to use SHA1 hashes of each article's normalized URL as the id for the article. Of course, if unique identifiers from the sources CMS are available, they might be used instead, ideally with a unique prefix: spiegel-187281.
articledata is purely a specification package, it does not contain any article processing code (e.g. crawling, NLP).
articledata is on pypi:
pip install articledataarticledata is used to move crawled news articles between these applications:
storyweb- an app to extract actor networks from corpora of news reporting.mediacrawl- a tool for crawling articles from a particular set of domains and extracting the article body.
See LICENSE, it's an MIT-style scheme.