Putting Things in Context: Situated Language Understanding for Human-Robot Dialog(ue)
Document Type Conference Paper
In Proceedings of Dialog with Robots: Papers from the AAAI 2010 Fall Symposium. Washington. Association for the Advancement of Artificial Intelligence.
Abstract
In this paper we present a model of language contextualization for spatially situated dialogue systems including service robots. The contextualization model addresses the problem of location sensitivity in language understanding for human-robot interaction. Our model is based on the application of situation-sensitive contextualization functions to a dialogue move's semantic roles -- both for the resolution of specified content and the augmentation of empty roles in cases of ellipsis. Unlike the previous use of default values, this methodology provides a context-dependent discourse process which reduces unnecessary artificial clarificatory statements. We detail this model and report on a number of user studies conducted with a simulated robotic system based on this model.