Query Understanding for Search on All Devices at WSDM 2016

It is our great pleasure to welcome you to the QRUMS 2016, the workshop on Query Understanding for Search on All Devices, held as part of the WSDM 2016 conference in San Francisco, USA. Theme and Purpose of the workshop: Query understanding has become a crucial component for today’s search engines. It is important to improve query understanding as query formulation is the way through which users express their search intent to a search engine. With the ubiquitousness of mobile devices, it is even more important to better understand user intent as typing on mobile is hard and time consuming. This workshop focuses on query understanding for mobile search with a motivation of reducing user efforts in formulating queries and getting their expected search results quickly. One of the most important problem in query understanding is Query auto-completion (QAC). QAC is the first service through which users interact with a search engine to input their search intent. QAC has to provide and update their suggestion lists based on each new character typed by the user in the search box. Returned suggestion lists are ranked based on different relevance models, such as most popular completion (based on historical frequency counts from query logs) [1], time-based (giving more weight to breaking news or recent popular queries) [10, 12, 11], location-based [9], context-sensitive (based on user’s context) [1, 5], personalized (based on user’s profile) [9], click modeling (based on user’s past clicks) [6], user-QAC interactions [8, 4], adaptive query auto-completion [13]. QAC becomes even more important as the focus shifts from desktop search to mobile search. On mobile, due to small typing keyboard, it is even more important to exploit all user and its context information available on mobile to provide the user better search suggestions. On query understanding and QAC: there is also a need of personalized and context models, especially for name queries. For example, a user may talk to his smart phone, “forward an email to Bill”. This implicitly triggers the query understanding engine to map “Bill” to a specific person in