Matrix Factorization in Latent Semantic Indexing

Abstract

Matrix factorizations are methods used to factorize a matrix into a product of two or more matrices. Matrix factorizations are used to reduce the dimension of a data set that help in reducing the computational time. In this project, we study how Singular Value Decomposition (SVD) and Non-Negative Matrix Factorization (NMF) are applied in Latent Semantic Indexing (LSI). LSI is a search algorithm where a set of documents is returned based on the keywords searched by the user. The performance of the two types of matrix factorizations are compared while applying them in LSI.