Department of Mathematics
"Overview of factor models applied to analysis of text documents"
Abstract: In this talk, I will go over factor models designed for the statistical analysis of text documents. A large text collection is often represented by a matrix indexed by its terms and documents. Each document corresponds to a column of this matrix and is a vector in the so-called term space. A factor model tries to find some commonality among terms and documents to reduce the dimension of term space. They involve lower-rank approximations of term-to-document matrix and produce dense lower-dimensional representations of documents. I will describe some of those models and illustrate them on numerical examples.
Date: Thursday, October 27, 2016
Place: Mathematics Seminar, SA-141
Tea and cookies will be served before the seminar.