Significant progress has been made in building  Question Answering systems that are focussed on providing precise answers to specific questions in one shot. In many real world situations, however, the information need may be specified vaguely and some sort of interaction may be required for a complete specification of the problem. Further, the correct answer of the question may be a procedure or passage rather than a factoid. For example, someone having trouble with battery life for her smartphone may express a query as ‘battery dying too soon for my XYZ phone’. The intent of the user here, obviously, is to seek a resolution to her problem. The resolution in question may be a procedure consisting of a sequence of steps. In order to recommend the correct resolution, a system may first have to engage in a dialog with the user to ascertain all the symptoms and match the correct resolution (‘answer’). 

In this workshop, we are looking to explore such advanced QA systems that seek to resolve more general user problems in an interactive manner. We are specifically interested in the problem of rapidly bootstrapping such QA systems with limited data annotation.