Learning to rank in production and user modeling for image search


This Friday we'll have two talks followed by drinks.

16:00 Diego Ceccarelli (Bloomberg London): Learning to Rank: From theory to production

Learning to Rank is awesome. Even more awesome is the fact that Apache Solr/Lucene is the first open source search engine that can do it out of the box. But all that is for nought if you don't hunt down the necessary features, make it inter-operate with all the other functionality, and do this fast enough on a production system for such ranking to be feasible.
This talk, by one of the engineers at Bloomberg who built this functionality into Solr in the first place, is a war story of how the company's real-time, low-latency news search engine was tamed to learn how to rank. Join us on a journey that will teach you how to take your LtR system to your clients, and more importantly, the many ways not to do it. There will be drama, excitement, and despair (and even Gandalf, if you pay attention)! Now grab that popcorn...


16:30 Xiaohui Xie (Tsinghua University): User behavior modeling for Web image search

Web-based image search engines differ from Web search engines greatly. The intents or goals behind human interactions with image search engines are different. In image search, users mainly search images instead of Web pages or online services. It is essential to know why people search for images because user satisfaction may vary as intent varies. Furthermore, image search engines show results differently. For example, grid-based placement is used in image search instead of the linear result list, so that users can browse result list both vertically and horizontally. Different user intents and system UIs lead to different user behavior. Thus, it is hard to apply standard user behavior models developed for general Web search to image search. In this talk, Xiaohui Xie will introduce two recent works on the intent taxonomy of image search users (accepted by WSDM 2018) and an interaction behavior model constructed to improve the ranking of image search engines (accepted by SIGIR 2018).