On the Relation Between "Semantically Tractable" Queries and AURA’s Question Formulation Facility

In our work on AURA, we are dealing with (controlled) natural language queries to a scientific knowledge base, and this paper explores the extent to which Popescu et al’s framework can be applied in the AURA context. The AURA task differs from the database task in several important ways, in particular the variety and complexity of questions is much more open-ended than in Popescu et als database work (Examples of some of the biology questions that AURA is intended to field are discussed in [3]). We briefly summarize Popescu et al’s work, and then explore the opportunities and challenges of applying it in a different context.