Solving for Satisfaction: Introduction to Click Models
Elizabeth Haubert • Location: Theater 5 • Back to Haystack 2019
Relevance metrics like NDGC or ERR require graded judgements to evaluate query relevance performance. But what happens when we don't know what 'good' looks like ahead of time? This talk will look at using click modeling techniques to infer relevance judgements from user interaction logs.
Elizabeth HaubertOpenSource Connections
Search relevance is all about marrying the nuts and bolts of data engineering with the art of constructing a user experience. In recent years, Elizabeth has worked with a spectrum of data transformation needs from high-rate, high-precision time-series sensor data to terabyte-scale text and image retrieval systems. Her current passion is setting up the analytics, infrastructure, and processes needed to bring machine learning to the tasks of search relevance.