It's Not You, it's Me: Automatically Extracting Social Meaning from Speed Dates
Speaker: Dan Jurafsky
Institution: Stanford University
Abstract:
A computational tool that can understand the meaning of words, sentences, and conversations is one of the oldest goals of artificial intelligence. In the 1970's and 1980's, such tools were built by hand, a laborious process which led many people to give up on robust understanding as a practical problem. In this talk I'll try to survey a number of attempts to learn both word meaning and world meaning automatically from text. I'll focus on the problem of relation extraction, models that learn "IS-A" and other semantic relation between words, but i'll also touch on work on ontology induction and on the induction of world knowledge structures like scripts.