Speaker: Sanne Vrijenhoek, Centrum Wiskunde en Informatica (CWI), the Netherlands
Moderator: Lynda Hardman, Centrum Wiskunde en Informatica (CWI), the Netherlands
The talk will also be live-streamed and recorded on the DIGHUM YouTube Channel.
Abstract
By accounting for individual preferences and differences between people, personalized news recommender systems can help readers find the news stories that are in some way relevant to them. In doing so, they are taking over the gatekeeping role traditionally held by human editors, and are thus in part shaping which news gets the public’s attention. Yet in their current form, news recommender systems primarily optimize for clicks, and do not or hardly account for the values that humans would adhere to when constructing an overview of the news.
For news recommender systems to reach their full potential, we need to translate abstract journalistic and editorial values, which are often ‘gut feelings’, into measurable technological artifacts. In this talk we will discuss the challenges of doing so, the work that has been done thus far, and look forward to the inter- and transdisciplinary issues that need to be resolved.