Title: Some Machine Learning Problems related to Content Optimization Speaker: Deepak Agarwal Affiliation: Yahoo! Labs I will discuss a relatively new problem of selecting the best content to show to a user who visits a site like Yahoo!, MSN, Digg etc. In particular, I will consider scenarios where the content pool to select from is dynamic with short article lifetimes. I will provide an in-depth discussion of modeling challenges (and some of our solutions) that arise in this scenario, viz, a) estimating click-through rates that exhibit both temporal and positional variations b) efficient explore/exploit schemes and c) personalization of content to individual users. Finally, I will end with discussion of some open problems in this area. Throughout, data from a content module published regularly on the Yahoo! Front Page will be used for illustration.