TopicScore: The Topic SCORE Algorithm to Fit Topic Models

Provides implementation of the "Topic SCORE" algorithm that is proposed by Tracy Ke and Minzhe Wang. The singular value decomposition step is optimized through the usage of svds() function in 'RSpectra' package, on a 'dgRMatrix' sparse matrix. Also provides a column-wise error measure in the word-topic matrix A, and an algorithm for recovering the topic-document matrix W given A and D based on quadratic programming. The details about the techniques are explained in the paper "A new SVD approach to optimal topic estimation" by Tracy Ke and Minzhe Wang (2017) <arXiv:1704.07016>.

Getting started

Package details

AuthorMinzhe Wang [aut, cre], Tracy Ke [aut]
MaintainerMinzhe Wang <minzhew@uchicago.edu>
LicenseMIT + file LICENSE
Version0.0.1
Package repositoryView on CRAN
Installation Install the latest version of this package by entering the following in R:
install.packages("TopicScore")

Try the TopicScore package in your browser

Any scripts or data that you put into this service are public.

TopicScore documentation built on June 6, 2019, 5:06 p.m.