I work on search evaluation datasets with large training data: MS MARCO(opens in new tab) and the TREC Deep Learning Track(opens in new tab). I’m interested in making search more interactive, such that the more you say about what you want, the better the results get: Neural Approaches to Conversational Information Retrieval(opens in new tab).

I am interested in Web search evaluation. I built the VLC, VLC2, WT2g and .GOV test collections, which have been made available to research groups around the world. I was involved in coordinating the TREC Web Track(opens in new tab) experiments that started in 1999, another round of TREC Web Track(opens in new tab) experiments starting in 2009 that introduced a diversity target to the evaluation, and most recently the TREC Deep Learning Track(opens in new tab) starting in 2019 that introduced large-scale training data for IR.

I also work on effective Web search, which includes the use of links and clicks to enhance the quality of search engine results, and more recently the use of representation learning and specifically neural information retrieval [FNTIR](opens in new tab).