Event Triples from News Articles

This is a collection of event triples that were extracted from English news outlets and news outlets that publish in the English language. For this collection, news articles for the time period  2016-01-01 to 2016-12-31 were retrieved. Entities are of type actor and location and are automatically annotated in the text of the news articles with the Stanford NER toolset; dates were annotated using Heideltime. Triples were constructed from co-occurrences of entities within a window of at most 3 consecutive sentences.

Information about the extracted news event triple is stored in the following format:

field description
e_id internal event ID
e_url news article url
e_geo_text text of extracted geo location
e_geo_id Wikidata ID of extracted geo location
e_geo_sent sentence number on news article
e_actor_text text of extracted actor
e_actor_id Wikidata ID of extracted actor
e_actor_sent sentence number on news article
e_time_text text of extracted time
e_time_norm normalized time
e_time_sent sentence number on news article

Information about the news outlets is given in the table below. The number of news articles for each outlet is given for the year 2016, even though some start dates are just within the year 2016 as new outlets are added continuously as part of the project.

Outlet Number of News Articles 2016 Country Start Date
Reuters 100349 England Jan, 2016
The Los Angeles Times 51071 USA May, 2016
The Washington Post 50674 USA May, 2016
Fox News 43770 USA May, 2016
The Sydney Morning Herald 31887 Australia May, 2016
The Independent 30405 England Jan, 2016
The Huffington Post – US 30049 USA May, 2016
The Telegraph 29248 England Jan, 2016
USA Today 23923 USA Jan, 2016
New York Times 23114 USA Jan, 2016
The Internat. Business Times 20178 USA Jan, 2016
The Guardian 19531 England Jan, 2016
The Associated Press 18133 USA May, 2016
CBS News 17291 USA May, 2016
The Huffington Post – UK 17014 England May, 2016
CNN 15642 USA May, 2016
BBC 9202 England Jan, 2016
Sky News 8773 England May, 2016
SBS 4084 Australia Nov, 2016
Deutsche Welle 2384 Germany Nov, 2016
Deutsche Welle 1989 Germany Nov, 2016
CBS News 1791 USA May, 2016
United Press International 1627 USA Nov, 2016
AlJazeera 1566 Catar Nov, 2016
New Yorker 1367 USA May, 2016
Die Basler Zeitung 969 Switzerland May, 2016
The Atlantic 264 Canada Nov, 2016
National Public Radio 84 USA Nov, 2016

The figure below shows the number of news article per month in the year 2016.

bildschirmfoto-2017-02-20-um-11-15-31

The following table gives the number of news article for the top 25 categories for the articles collected for the year 2016.

Category Number of Articles
world 80558
business 72636
politics 63902
companyNews 61861
finance 45068
newsWorld 39291
national 34455
newsUK 21031
political 18057
money 16026
news 12626
US 12295
local 8357
washington 8118
newsUS 6942
national(US) 6565
worldnews 6230
USNational 5874
politicsUS 4754
technology 3986
environment 2996
europe 2940
all 2800
internat. Business 2607
UK 2538

Download

Because of the large size of the whole News Event Triples dataset, we also provide the triples in ten separate files. There is no particular ordering or organization upon which the partitioning of the complete dataset was conducted:

Event_News_Triples_all.7z
(129 MB compressed, 27 GB uncompressed, 172.521.530 event instances)

Separate files:

Event_News_Triples1v10.7z
(23 MB compressed, 2.6 GB uncompressed, 17.252.153 event instances)

Event_News_Triples2v10.7z
(10 MB compressed, 2.5 GB uncompressed, 17.252.153 event instances)

Event_News_Triples3v10.7z
(9 MB compressed, 2.5 GB uncompressed, 17.252.153 event instances)

Event_News_Triples4v10.7z
(11 MB compressed, 2.5 GB uncompressed, 17.252.153 event instances)

Event_News_Triples5v10.7z
(10 MB compressed, 2.6 GB uncompressed, 17.252.153 event instances)

Event_News_Triples6v10.7z
(9 MB compressed, 2.7 GB uncompressed, 17.252.153 event instances)

Event_News_Triples7v10.7z
(14 MB compressed, 2.7 GB uncompressed, 17.252.153 event instances)

Event_News_Triples8v10.7z
(14 MB compressed, 2.7 GB uncompressed, 17.252.153 event instances)

Event_News_Triples9v10.7z
(12 MB compressed, 2.8 GB uncompressed, 17.252.153 event instances)

Event_News_Triples10v10.7z
(18 MB compressed, 3.2 GB uncompressed, 17.252.153 event instances)