Peak Preview: Cross-language news investigations

It is no secret that news travels around the world, crossing borders and languages in order to reach its audience. That’s why one important aspect of any digital news analysis or investigation is making sure that language does not become a barrier, but rather a facilitator.

As we are preparing for our TrustServista version 1.0 launch (20th of June), we wanted to share with you one of our most exciting features, still being tested in our data science lab for a future release later this year: a cross-language investigation that spans across 4 languages: Russian, English, French and German.

The starting point of this TrustServista exercise is an article from Russian news agency RIA Novosti: На Ближнем Востоке Трамп ищет друзей против Ирана. Мешает “русский скандал”. This article references 6 other articles, from English, French and German websites located in Israel (The Jerusalem Post), Syria (Al-Masdar), France and Germany:

In order to normalize the language and be able to calculate the TrustLevel an determine PatientZero, the non-English articles were first translated automatically into English using SDL technologiesThe translated output was then published on our demonstration website for persistence purposes and then – starting from the original Russian article from (now in English), the links to other articles were built, named entities extracted and the key metrics of TrustServista – the TrustLevel and PatientZero – were determined.

Already have a TrustServista account?

Analyze Multilanguage Article

If you don’t have an account, you care request one here.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.