Learning to rank has emerged as an important problem at the intersection of machine learning and information retrieval. Driven by applications in web search, researchers have sought to build learning systems that can automatically rank documents in order of their relevance to a search query. Despite the tremendous amount of activity in this area during the past decade, there are still many theoretical issues that researchers have just begun addressing. In this talk I will present some recent advances in deriving generalization error bounds and asymptotic consistency guarantees for learning to rank algorithms. Time permitting, I will also talk about emerging topics in learning to rank such as online learning and learning with user feedback only on the top few documents.
This talk is based on collaborative work with Shivani Agarwal (IISc Bangalore),Sougata Chaudhuri (Univ. of Michigan), Harish Ramaswamy (IISc Bangalore),Pradeep Ravikumar (UT Austin) and Eunho Yang (IBM TJ Watson).