Deepnews, solving real problems

Don’t fret, I’m not going to make a post a day on startups that are « solving real problems ». But I thought it would be fair to point at Deepnews, an interesting European startup (it’s even French-based — oh my). Deepnews is solving a dramatically obvious problem: news that have journalistic value and that « add to the conversation » are rarer and rarer. In 2020, this type of news drowns in the immense background noise of social media, click-baits, propaganda bots, and just irrelevant nonsense.

The key idea here is to use a deep learning-based algorithm to screen through the daily ocean of non-sense:

The Deepnews Scoring Model (DSM) works like this: a story gets passed through it and receives a quality score on a scale of 1 to 5. The score reflects the journalistic added-value: depth of reporting, variety of sources, precision, writing structure, etc. Frédéric FILLOUX

And yes, the irony of calling the service deepnews in regard to the expected explosion of deepfakes is not lost on you.