top of page

Intention-Aware Recommendations and User Satisfaction

Cx researchers are exploring ways to make Internet recommendations more helpful and less harmful. The algorithms that recommend news, music, videos, and other content tend to fall into the trap of optimizing for clicks or other easy-to-measure forms of engagement, giving people popular content rather than content that they actually aspire to consume. The research, which was also covered in Wired, offers ways to think about increasing engagement while also ensuring that users receive satisfying, high-quality content.

Cx PhD student Longqi Yang and colleagues at Cornell Tech, CUNY, and Himalaya Media conducted a four-week field study to investigate the effects of recommending content based on users' explicitly stated interests rather than by popularity. Participants in the study downloaded a modified version of the Himalaya podcast recommendation app that asked them about the content they aspired to listen to, and subscribe to channels based on those aspirations. They then varied the type of recommendation - aspirational vs popularity-based - that they received. Participants also either got recommendations that were just based on the channels they had subscribed to or a combination of subscription and newly recommended content.

The results showed that users who were shown aspirational content interacted with the content more, and exploratory recommendations - those that included channels to which a user wasn't already subscribed - improved user satisfaction with the app, but only for those who were already receiving intention-aware recommendations. The results suggest that using information about what users intend to listen to, rather than what's popular overall, can both facilitate content consumption and provide a more satisfying user experience overall.

The research will be presented at The Web Conference 2019 in San Francisco, CA. You can find an advance copy of the full paper here.


Recent Posts

See All

Cx PhD student Yiqing Hua has been working on using natural language processing techniques to detect propaganda in news. As a participant in the 2nd Workshop on NLP for Internet Freedom (NLP4IF), Y

bottom of page