Augmenting the power of the various versions of LSI used in document retrieval

Authors:
Hua Yan;William I. Grosky;Farshad Fotouhi
Affiliations:
Computer Science Department, Wayne State University, Detroit, MI;Department of Computer and Information Science, University of Michigan-Dearborn, Dearborn, MI;Computer Science Department, Wayne State University, Detroit, MI
Venue:
DNIS'05 Proceedings of the 4th international conference on Databases in Networked Information Systems
Year:
2005

Citing 4
Cited 0

Using linear algebra for intelligent information retrieval

SIAM Review
A semidiscrete matrix decomposition for latent semantic indexing information retrieval

ACM Transactions on Information Systems (TOIS)
Latent semantic space: iterative scaling improves precision of inter-document similarity measurement

SIGIR '00 Proceedings of the 23rd annual international ACM SIGIR conference on Research and development in information retrieval
Introduction to Modern Information Retrieval

Introduction to Modern Information Retrieval

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper clarifies the apparent randomness and confusion in the choice of different query methods using LSI to be found in the current text and image retrieval literature. We also propose and test some modified query approaches using the novel technique of singular value rescaling (SVR). Experiments on standardized TREC data set confirmed the effectiveness of SVR, showing an improvement ratio of 5.6% over the best conventional LSI query approach.