Pragmatically similar utterance finder demonstration

Models for estimating the similarity between two utterances are fundamental in speech technology. While fairly good automatic measures exist for semantic similarity, we only recently built a model of pragmatic similarity, the first. We propose to present this model by letting participants try out our Pragmatically Similar Utterance Finder. This system listens to one side of a live conversation, identifies the utterances, and for each retrieves the most similar utterances, according to our model, from a large corpus. Participants and viewers will then be able to hear these utterances, and judge for themselves the prospects for pragmatic-similarity modeling.