Online media / Recommend

Online Media

Challenge

In increasingly competitive environment, a major German online news provider needed to find ways to keep users engaged. They suspected strong recommendations for articles based on users’ behavior might be the key. A large team of data scientists had already worked on the problem for months and substantially improved clickthrough rates, but they were interested in trying new approaches to get even better results. The challenge: Spread across the sites, the customer gathers over 100 GB of data from over 5 million unique users every day.

Solution

In a short, four-week project, Curiosity is developing a recommender engine based on a graph of the information about all the customers’ articles and users. It works by automatically extracting topics from each incoming news article and measuring its similarity to articles and topics that users have viewed in the past. When users click on an article, it will use that knowledge to recommend articles to them. When it’s ready, it will handle the full volume of the production system and return recommendations in milliseconds.

Products
  • Date: June 2018

contact

Schedule a demo

Thanks for contacting us, we'll be in touch soon!

Office

Werk1, Grafingerstr. 6, 81671 Munich, Germany

Contact

+49 151 1766 7073
hello@curiosity.ai

LinkedIn

Twitter

Medium

© 2018 Curiosity GmbH. All Rights Reserved. Legal Notice