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
    • General Computer Science

    Fingerprint

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

    Cite this