D 2011

Subspace Tracking for Latent Semantic Analysis

ŘEHŮŘEK, Radim

Basic information

Original name

Subspace Tracking for Latent Semantic Analysis

Authors

ŘEHŮŘEK, Radim (203 Czech Republic, guarantor, belonging to the institution)

Edition

Heidelberg, Proceedings of the 33rd European Conference on Information Retrieval (ECIR), p. 289-300, 12 pp. 2011

Publisher

Springer

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10000 1. Natural Sciences

Country of publisher

Ireland

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

printed version "print"

References:

Impact factor

Impact factor: 0.402 in 2005

RIV identification code

RIV/00216224:14330/11:00067252

Organization unit

Faculty of Informatics

ISBN

978-3-642-20160-8

ISSN

DOI

UT WoS

000301968000029

Keywords in English

scalability svd subspace tracking

Tags

International impact, Reviewed
Změněno: 30/4/2014 04:36, RNDr. Pavel Šmerk, Ph.D.

Abstract

V originále

Modern applications of Latent Semantic Analysis (LSA) must deal with enormous (often practically infinite) data collections, calling for a single-pass matrix decomposition algorithm that operates in constant memory w.r.t. the collection size. This paper introduces a \emph{streamed distributed algorithm for incremental SVD updates}. Apart from the theoretical derivation, we present experiments measuring numerical accuracy and runtime performance of the algorithm over several data collections, one of which is the whole of the English Wikipedia.

Links

LC536, research and development project
Name: Centrum komputační lingvistiky
Investor: Ministry of Education, Youth and Sports of the CR, Centrum komputační lingvistiky