Online Learning to Rank
Textkernel Tech Talks, Amsterdam, The Netherlands. Jun 26, 2014.
Summary
This presentation explores online learning to rank methodologies in information retrieval systems, addressing the limitations of traditional offline evaluation approaches that rely on fixed optimal rankings and pooled relevance assessments. The work examines how search engines can dynamically learn and adapt ranking functions in real-time user interactions, moving beyond static evaluation metrics like precision and recall to incorporate actual user behavior and satisfaction. The research contributes to improving search relevance in practical applications like job/candidate matching by developing learning systems that can adapt to user preferences and overcome the inherent biases present in offline evaluation methodologies.
Slides
Links
Related Publications
Search Engines that Learn from Their Users
Anne Schuth.
2016.