On April 25, 40 first-place recipients were honoured during the 2024 Global Media Awards in London. One of them was MediaFutures partner Bergens Tidende (BT) in cooperation with associate professor Mehdi Elahi and UiB student Peter Kolbeinsen Klingenberg.
Their “Personalised News Experience” won the first price in the category Readership and Engagement, regional brands.
“Of course I am excited about this. It was a great collaboration between us in MediaFuture and BT as our industry partner”, says professor Mehdi Elahi and adds:
“I believe, it was well deserved. The findings of the thesis showed effectiveness of the recommendation approach.”
Using content- and behavioural data for recommendations in the Norwegian news market
The project started with Peter’s master’s thesis, where he looked into using data about what people read and how they behave to suggest news articles.
In his thesis, he focused on two recommendation techniques. One is collaborative filtering which generates recommendations based on what what people with similar interests read, while the other looks at the content of the articles themselves.
“We tweaked and experimented a lot with how we recommend articles on bt.no, which may be less appealing than a very concrete project to improve reader engagement. Our nomination was actually a combination of two several projects: one internal project for frontpage optimization, and another project on optimization for increasing article recirculation (this project was part of the MediaFutures collaboration)”, says data scientist in BT, Thomas Husken.
Both methods seemed to help get more people to click on articles, but focusing on the article content stood out as the best.
By using a special language model called sBERT, made by the National Library’s AI Lab, they could compare articles and suggest the top 5 most similar ones for people to read more.
Using this method, they saw a big 20% increase in the number of clicks compared to the old system Bergens Tidende used.
“The award definitely came like a surprise, for several reasons. I didn’t expect to receive an award for our work that is a bit more abstract in nature than some other nominees.”, Husken says and continues:
“Besides AI, personalisation is the one thing that every news brand in the world is working on in some way. To be recognized for that in a global award, is absolutely amazing.”
Photo taken by Anna Pacholczyk: Associate professor Mehdi Elahi and Thomas Husken from Bergens Tidende supervised Peters master thesis.
We want to congratulate Peter Kolbeinsen Klingenberg, Associate Professor Mehdi Elahi and Thomas Husken for achieving such a commendable outcome through their professional collaboration.