News personae and the future of AI recommendations

Personalisation is a developing domain in news publishing. 67% of publishers are already using AI personalisation recommenders, but what’s next? A recent research paper from the KU Leuven brings a new perspective in the interaction between publishers, readers and algorithms. The paper “Putting a Human Face on the Algorithm: Co-Designing Recommender Personae to Democratize News Recommender Systems” details the concept of news recommender personae. Twipe’s Riccardo Fredro talked to the authors of the paper, Lawrence Van den Bogaert and Jaron Harambam to understand why they think this can play a role in the future of news. Watch the full conversation here.

What are news personae?

AI news personae use multiple algorithms per user to tailor recommendations on a 1-to-1 basis for different reading situations. When visiting a website, readers get presented with the possibility to read 3 different curated feeds rather than 1. The approach is explained in more detail in the video below.

3 personae to cover different reading demands

To create the news personae, the research team studied the news consumption habits of a control group of 15 news readers. Each member kept a diary, noting down which articles they had read and providing a reflection of 1 article per week. From this data, the researchers distilled a set of core reading motivations which were made into 8 persona.

These 8 persona were presented to a new control group of 12 people with a fresh perspective who were asked to fit each persona to a specific news story. 8 persona then became 3, with the decision made that people shouldn’t have to identify with personae on a 1 to 1 level. The final personae were:

  • The challenger: A persona designed to challenge your pre-existing beliefs and present alternative viewpoints.
  • The expert: Takes readers into deeper detail (including purposeful filter bubbles) and maximises the level of knowledge gained from their reading experience.
  • The unwinder: Presents readers with lighter articles on topics of existing interest and non-news topics.

Each persona was then given a visual identity to add to the transparency of the algorithms and make them approachable. The news personae algorithms give the end user greater agency around the recommendations that they receive. Different persona can be chosen at different points in the day to match their reading mood and needs. Aside from the 3 personae, readers can also still access the news without any element of personalisation.

How publishers have been leveraging personae in product development

Publishers have adopted forms of personae in product development to engage new audiences. When designing l’Édition du Soir, Ouest-France focused on the persona of the unwinder. The evening edition product was designed after studying an evening increase in user behaviour and noticing peaks on content like quizzes and games. L’Édition du Soir was created with these readers and their interests at heart, including videos, games, puzzles and also short stories about the main news. The New York Times have also leveraged this persona with their bundle products offering readers more than just the news.

The expert persona has been leveraged in niche publications. The Athletic have successfully adopted this model in their reinvention of local sports coverage. Hiring specific reporters and data experts to focus on specific teams and sports, they have popularised the long-read and deep dive concepts for sports readers. Their coverage is designed to bring sports fans closer to the heart of the game. Ahead of 2022, The Athletic even launched an ”emergency football hotline” to provide fans with expert opinions.

Back in 2018, The Guardian experimented with a commuter persona through their app LabRDR. LabRDR was an offline experiment where users got ‘packages’ of Guardian articles delivered to their phones twice a day. These packages were sent at times they determined to fit their commute. The packages were tailored to the length of consumer’s commutes and were personalised to provide a set of stories the commuter would want to read. March 2018 was the last time the app was updated, with the product then discontinued.

JAMES news recommendations are leveraged by publishers like NRC and Keskisuomalainen and act as a form of “explorer” persona. The personalised recommendations use a hybrid algorithm to combine trending news with personalised interest areas. In doing so, readers are typically exposed to 80-90% of the day’s news content, being given a broader depth of news.

What can publishers learn from personae?

Whilst the news personae algorithm concept is not yet implemented, it does stimulate certain discussion points for product and data teams. We’ve listed a selection of ours below:

  • Many publishers have not yet got 1 recommender algorithm running frequently, let alone 3. JAMES personalisation could help you get on your way.
  • The news recommender personae concept is a good chance for publishers to look to the future and get ahead of big tech.
  • Multi-title news publishers could use these reading personae to expose readers to content away from their chosen title and push readers into expanding their subscription.
  • News personae could be the perfect partner in life tool to accompany your readers through their reading experience. In a fight for retention, they could make the difference.
  • Whilst offering personalised news experiences to readers, it is important to give readers the choice of personalisation or editorially curated experiences. Trust in the news is low and this transparency is a chance to rebuild strained relationships.
Matthew Lynes
Media Innovation Analyst @ Twipe

Subscribe to our Future of News newsletter

Get weekly insights on digital publishing, artificial intelligence, and paid content straight in your mailbox.

 

 

*We’re committed to your privacy. Twipe might use the information you provide to contact you about relevant content, events, products and services. You may unsubscribe from these communications at any time. For more information, check out our privacy policy