Recsperts - Recommender Systems Experts #21: User-Centric Evaluation and Interactive Recommender Systems with Martijn Willemsen
Apr 8, 2024
Martijn Willemsen, expert in interactive recommender systems, discusses empowering users with control over recommendations, understanding user goals for better satisfaction, and the psychology of decision-making in recommendation systems. They explore music recommender systems, nudging users towards new genres, and the value of user feedback for improved recommendations.
AI Snips
Chapters
Transcript
Episode notes
Clicks vs. Satisfaction
- In a video recommender study, users clicked less with better recommendations.
- They achieved their goal faster, demonstrating clicks don't always mean engagement.
Implicit Data and Goals
- Implicit data like clicks doesn't reveal user goals.
- Recommending based on past behavior can be suboptimal for achieving user goals.
Goal-Oriented Recommenders
- Understand user needs and goals through user research.
- Build recommender systems that support users diverse goals, not just past likes.