One Step Forward and Two Steps Back in Conversational Models
Abstract
With conversational robotics on the rise in applications from Siri to Jibo, computational models of language are becoming ever more present in our lives. In this talk, I discuss my experience in development of a planning-based dialogue system implemented in a robotic architecture. I show that there are fundamental aspects of language with which we must grapple for robust language understanding. I demonstrate the tooling I've developed to examine human dialogue, and how I am currently using that tooling to enhance our understanding of natural language use in humans. Finally, I illustrate a research path for applying the ongoing basic research into enhanced dialogue systems.
Please join meeting in Sococo VH 209. Login: tuftscs.sococo.com
Join Zoom meeting: https://tufts.zoom.us/j/94421971324
Password: see colloquium email
Dial-in option not available.