Principal Investigator: Mirella Lapata

The overall aim of this fellowship is to develop novel algorithms and techniques for text rewriting for a number of key text-to-text generation applications, and through this to then develop a general modelling framework of wide applicability for capturing alternative ways of conveying the same information at the sentence and document level. We will do this by adopting a data intensive approach where meaning equivalences will be learned directly from text samples by exploiting what is observable in the data rather than manually specified rules, or elaborate semantic representations.