Recomputation of class relevance scores for improving text classification

Sang Bum Kim, Hae Chang Rim

Research output: Chapter in Book/Report/Conference proceedingChapter

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 languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
EditorsAlexander Gelbukh
PublisherSpringer Verlag
Pages580-583
Number of pages4
ISBN (Print)3540210067, 9783540210061
DOIs
Publication statusPublished - 2004

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2945
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Recomputation of class relevance scores for improving text classification'. Together they form a unique fingerprint.

Cite this