Vector representations of sentences and other large units
September 16 - 18, 2015, We/Th/Fr
Organizers: Stephen Clark and Georgiana Dinu
Original Topic Description: Vector-based sentence representations do well on many tasks: paraphrasing, simple question answering, and even basic machine translation. Thus, the question arises: can vector-based sentence representations be extended to be a general model of sentence meaning? Or are there semantic problems vector-based sentence representations cannot address? Natural language generation is one interesting problem in this context: is it possible to do generation from vector representations?