Do you like that t-shirt? Then you might be interested in these jeans too. Companies like Amazon and Google are well known for their evaluation and recommendation methodologies. In order for universities to test their algorithms on the scale of such commercial parties, NewsREEL organized a global contest and.. ORTEC won the first prize!
Four ORTEC employees of the cross-business unit Amsterdam Smart Team decided to join the competition, named Pieter Bons, Nick Evans, Peter Kampstra and Timo van Kessel. Creating algorithms for ORTEC’s Internal Communications platform is their daily business. Moreover, ORTEC’s platform works similar to Amazon and Google: after reading an article, employees can receive recommendations for articles they might like as well. In this way, employees are encouraged to read more internal news.
Using storytelling to make people click
The team’s goal was to test ORTEC’s “Killer Sequence algorithm” and find out how ORTEC measured up to the other teams. Pieter Bons: “We expected a strong correlation between the topic of a current article and the next click. Once you have read a sports article, your likelihood of clicking another sports article has increased a lot. The idea is that people will follow a certain path of storytelling, so we compared a sequence of articles to the sequences of other people.”
After distributing thousands of article recommendations, the team observed whether people clicked on the recommended articles. This was tested on news websites in a real environment, such as on www.bz-berlin.de. And the results don’t lie. Nick Evans: “The ORTEC algorithm scored a CTR of 2.09%, clearly defeating the next best team with a CTR of 1.57%. I am very proud of our team!”
Want to hear more about the project?
Last week, the team travelled to Dublin to present their research paper at the NewsREEL Conference. If you would like to find out how the recommender engine could work for your company, let us know. We’re always up for a cup of coffee!
Most recent articles: