Title: Scalable, General-purpose Semantic Learning 

Abstract:

In this talk, I will present recent work on two semantic tasks: (1)
all-of-vocabulary sense distribution learning, where we show that topic models
can be used to learn sense distribution data at a level comparable to or
better than SemCor, but over a much larger vocabulary; and (2) open-class
lexical relation classification, where we explore the general utility of
vector differences over word embeddings to capture the relation between
ordered word tuples.