Building Event-Centric Knowledge Graphs From News

[David Caswell]

The three-year, $4 million euro, 12 university EU NewsReader project completed on January 1st, 2016.

For those who are not familiar with the project, it’s a fully-automated process that ingests hundreds of thousands of text articles, identifies and individuates discrete news events mentioned within those articles, structures those news events and then stores them as a knowledge graph of structured events. It is a machine that reads general news and converts it into structured data.

A relatively non-technical paper describing the system will be published in the next issue of the Journal of Web Semantics - it’s temporarily available here: https://dkm-static.fbk.eu/people/rospocher/files/pubs/2016jws.pdf

There are very few concepts that one could imagine as a plausible alternative to a news ecosystem based on text articles - these Event-Centric Knowledge Graphs are one of them. My personal view is that fully-automated ECKG systems won't work in production, for a variety of reasons, but that editorially-curated ECKGs (such as my own Structured Stories) are potentially a major opportunity for journalism.

[David Caswell]

Easy-to-understand 4-minute video explaining NewsReader: https://youtu.be/rYLaVN3oqLI