The interactive transcript could not be loaded.



Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Feb 15, 2016

Recording history by reading massive streams of news. This is Sheila, a senior spokesperson of the ministry of internal affairs. She monitors the daily stream of news: hundreds of documents per day. When Sheila reads an article with the rumor of falsifications in CO2 emissions by a national automobile manufacturer, she needs to advise her minister on a response in the media within 60 minutes. To do this well, she needs an accurate picture of the full history, spanning decades of the manufacturer, the management currently in charge and the connections with the government.

Access to an overwhelming amount of data, millions of relevant documents, makes it almost impossible to find for Sheila the information she needs. Like Sheila, professionals in any sector depend on access to accurate and complete knowledge to make well informed decisions. Think about lawyers, politicians, head of purchase in a large firm, compliance directors and journalists. A missing piece of crucial information can be fatal for making the right decision. To find the piece, they need to search for a needle in a haystack. Simply because there is more data than ever. It is highly interconnected through the internet and quickly gets out of date in this rapidly changing world.

This is where Newsreader comes in. A reading machine, a machine that can read millions of news items in four languages every day. Newsreader allows you to find that needle in the haystack by structuring information as stories relating new events to past events. It stores all the details in a knowledge store and can intelligently reason over it by reconstructing histories over decades. It is therefore able to tell all the details about any person or organization as told by thousands of different sources. And it can measure the implications of an event. It measures not how much talk there is about what happened but how much actually happened. Who, for example, worked when for what company. Who owned what when or whether sales went up or down.

How does this work? In four simple steps: identification, deduplication, aggregation and perspectivation. I.D.A.P.
1) Newsreader first identifies an event in text, through similar components. By extracting what happened to whom, when and where.
2) Then Newsreader makes sure similar information across many sources in the haystack is represented only once, referencing every article.
3) Newsreader aggregates complimentary information across thousands of different sources in a single representation.
4) Makes sure differences and different viewpoints are traceable through their sources and mentions in text.

The result of this I.D.A.P. method is a complete, exact and rich record of the past. With access to original sources, the information from the news is stored as millions of factual statements, so called triples, representing what happened. With billions of archived documents and millions of documents added every day, we need to be able to effectively search this pile of information if we want to use the available information well. Newsreader offers a solution by reading billions of documents into an unfolding history with information about what happened to whom, when and where. It allows Sheila and professional decision makers like her, to make a well informed decision, based on accurate and complete knowledge, despite the overload of information available.


Dit filmpje is gemaakt door Bord&Stift. Wij helpen jou om jouw verhaal te vertellen met een whiteboard en stiften. Zo toveren wij complexe verhalen om tot simpele getekende filmpjes. Zie ook www.bordenstift.nl.


When autoplay is enabled, a suggested video will automatically play next.

Up next

to add this to Watch Later

Add to

Loading playlists...