Adaptive Search Engines • Current Search Engines

– Hand-tuned retrieval function • Hypothesis – Different users need different retrieval functions [Teevan et al. 07] – Different collections need different retrieval functions • Machine Learning – Learn improved retrieval functions – User Feedback as training data Overview of Talk • Understanding how users act – User study of Web search behavior using eye-tracking • How clicks relate to relevance – Interpreting clicks as relative vs. absolute feedback – Dealing with presentation bias – Accuracy of feedback strategies • Learning from user behavior – Learning ranking functions: Ranking SV