Recomputation of class relevance scores for improving text classification

Sang Bum Kim; Hae Chang Rim

doi:10.1007/978-3-540-24630-5_71

Recomputation of class relevance scores for improving text classification

Sang Bum Kim, Hae Chang Rim

* Honorary professors

Research output: Chapter in Book/Report/Conference proceeding › Chapter

Abstract

In the text classification task, bag-of-word representation causes a critical problem when the prediction powers for a few words are estimated terribly inaccurately because of the lack of the training documents. In this paper, we propose recomputation of class relenvace scores based on the similarities among the classes for improving text classification. Through the experiments using two different baseline classifiers and two different test data, we prove that our proposed method consistently outperforms the traditional text classification strategy.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Editors	Alexander Gelbukh
Publisher	Springer Verlag
Pages	580-583
Number of pages	4
ISBN (Print)	3540210067, 9783540210061
DOIs	https://doi.org/10.1007/978-3-540-24630-5_71
Publication status	Published - 2004

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	2945
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-540-24630-5_71

Cite this

Kim, S. B., & Rim, H. C. (2004). Recomputation of class relevance scores for improving text classification. In A. Gelbukh (Ed.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 580-583). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2945). Springer Verlag. https://doi.org/10.1007/978-3-540-24630-5_71

Recomputation of class relevance scores for improving text classification. / Kim, Sang Bum; Rim, Hae Chang.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). ed. / Alexander Gelbukh. Springer Verlag, 2004. p. 580-583 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2945).

Research output: Chapter in Book/Report/Conference proceeding › Chapter

Kim, SB & Rim, HC 2004, Recomputation of class relevance scores for improving text classification. in A Gelbukh (ed.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 2945, Springer Verlag, pp. 580-583. https://doi.org/10.1007/978-3-540-24630-5_71

Kim SB, Rim HC. Recomputation of class relevance scores for improving text classification. In Gelbukh A, editor, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Springer Verlag. 2004. p. 580-583. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-540-24630-5_71

Kim, Sang Bum ; Rim, Hae Chang. / Recomputation of class relevance scores for improving text classification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). editor / Alexander Gelbukh. Springer Verlag, 2004. pp. 580-583 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inbook{aa7ae00cd043456d86ca1e14b626e1c8,

title = "Recomputation of class relevance scores for improving text classification",

abstract = "In the text classification task, bag-of-word representation causes a critical problem when the prediction powers for a few words are estimated terribly inaccurately because of the lack of the training documents. In this paper, we propose recomputation of class relenvace scores based on the similarities among the classes for improving text classification. Through the experiments using two different baseline classifiers and two different test data, we prove that our proposed method consistently outperforms the traditional text classification strategy.",

author = "Kim, {Sang Bum} and Rim, {Hae Chang}",

year = "2004",

doi = "10.1007/978-3-540-24630-5_71",

language = "English",

isbn = "3540210067",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "580--583",

editor = "Alexander Gelbukh",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - CHAP

T1 - Recomputation of class relevance scores for improving text classification

AU - Kim, Sang Bum

AU - Rim, Hae Chang

PY - 2004

Y1 - 2004

N2 - In the text classification task, bag-of-word representation causes a critical problem when the prediction powers for a few words are estimated terribly inaccurately because of the lack of the training documents. In this paper, we propose recomputation of class relenvace scores based on the similarities among the classes for improving text classification. Through the experiments using two different baseline classifiers and two different test data, we prove that our proposed method consistently outperforms the traditional text classification strategy.

AB - In the text classification task, bag-of-word representation causes a critical problem when the prediction powers for a few words are estimated terribly inaccurately because of the lack of the training documents. In this paper, we propose recomputation of class relenvace scores based on the similarities among the classes for improving text classification. Through the experiments using two different baseline classifiers and two different test data, we prove that our proposed method consistently outperforms the traditional text classification strategy.

UR - http://www.scopus.com/inward/record.url?scp=35048853276&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=35048853276&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-24630-5_71

DO - 10.1007/978-3-540-24630-5_71

M3 - Chapter

AN - SCOPUS:35048853276

SN - 3540210067

SN - 9783540210061

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 580

EP - 583

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

A2 - Gelbukh, Alexander

PB - Springer Verlag

ER -

Recomputation of class relevance scores for improving text classification

Abstract

Publication series

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this