Some effective techniques for naive bayes text classification

Sang Bum Kim; Kyoung Soo Han; Hae Chang Rim

doi:10.1109/TKDE.2006.180

Some effective techniques for naive bayes text classification

Sang Bum Kim, Kyoung Soo Han, Hae Chang Rim

* Honorary professors

Research output: Contribution to journal › Article › peer-review

399 Citations (Scopus)

Abstract

While naive Bayes is quite effective in various data mining tasks, it shows a disappointing result in the automatic text classification problem. Based on the observation of naive Bayes for the natural language text, we found a serious problem in the parameter estimation process, which causes poor results in text classification domain. In this paper, we propose two empirical heuristics: per-document text normalization and feature weighting method. While these are somewhat ad hoc methods, our proposed naive Bayes text classifier performs very well in the standard benchmark collections, competing with state-of-the-art text classifiers based on a highly complex learning method such as SVM.

Original language	English
Pages (from-to)	1457-1466
Number of pages	10
Journal	IEEE Transactions on Knowledge and Data Engineering
Volume	18
Issue number	11
DOIs	https://doi.org/10.1109/TKDE.2006.180
Publication status	Published - 2006 Nov

Bibliographical note

Funding Information:
This work was partly supported by the JSPS Postdoctoral Fellowship Program and the Okumura Group at Tokyo Institute of Technology. H.-C. Rim was the corresponding author.

Keywords

Poisson model
Text classification
feature weighting
naive Bayes classifier

ASJC Scopus subject areas

Information Systems
Computer Science Applications
Computational Theory and Mathematics

Access to Document

10.1109/TKDE.2006.180

Cite this

@article{e81c99201acb4d828bbd2244625dc263,

title = "Some effective techniques for naive bayes text classification",

abstract = "While naive Bayes is quite effective in various data mining tasks, it shows a disappointing result in the automatic text classification problem. Based on the observation of naive Bayes for the natural language text, we found a serious problem in the parameter estimation process, which causes poor results in text classification domain. In this paper, we propose two empirical heuristics: per-document text normalization and feature weighting method. While these are somewhat ad hoc methods, our proposed naive Bayes text classifier performs very well in the standard benchmark collections, competing with state-of-the-art text classifiers based on a highly complex learning method such as SVM.",

keywords = "Poisson model, Text classification, feature weighting, naive Bayes classifier",

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

note = "Funding Information: This work was partly supported by the JSPS Postdoctoral Fellowship Program and the Okumura Group at Tokyo Institute of Technology. H.-C. Rim was the corresponding author.",

year = "2006",

month = nov,

doi = "10.1109/TKDE.2006.180",

language = "English",

volume = "18",

pages = "1457--1466",

journal = "IEEE Transactions on Knowledge and Data Engineering",

issn = "1041-4347",

publisher = "IEEE Computer Society",

number = "11",

}

TY - JOUR

T1 - Some effective techniques for naive bayes text classification

AU - Kim, Sang Bum

AU - Han, Kyoung Soo

AU - Rim, Hae Chang

N1 - Funding Information: This work was partly supported by the JSPS Postdoctoral Fellowship Program and the Okumura Group at Tokyo Institute of Technology. H.-C. Rim was the corresponding author.

PY - 2006/11

Y1 - 2006/11

N2 - While naive Bayes is quite effective in various data mining tasks, it shows a disappointing result in the automatic text classification problem. Based on the observation of naive Bayes for the natural language text, we found a serious problem in the parameter estimation process, which causes poor results in text classification domain. In this paper, we propose two empirical heuristics: per-document text normalization and feature weighting method. While these are somewhat ad hoc methods, our proposed naive Bayes text classifier performs very well in the standard benchmark collections, competing with state-of-the-art text classifiers based on a highly complex learning method such as SVM.

AB - While naive Bayes is quite effective in various data mining tasks, it shows a disappointing result in the automatic text classification problem. Based on the observation of naive Bayes for the natural language text, we found a serious problem in the parameter estimation process, which causes poor results in text classification domain. In this paper, we propose two empirical heuristics: per-document text normalization and feature weighting method. While these are somewhat ad hoc methods, our proposed naive Bayes text classifier performs very well in the standard benchmark collections, competing with state-of-the-art text classifiers based on a highly complex learning method such as SVM.

KW - Poisson model

KW - Text classification

KW - feature weighting

KW - naive Bayes classifier

UR - http://www.scopus.com/inward/record.url?scp=85008014387&partnerID=8YFLogxK

U2 - 10.1109/TKDE.2006.180

DO - 10.1109/TKDE.2006.180

M3 - Article

AN - SCOPUS:85008014387

SN - 1041-4347

VL - 18

SP - 1457

EP - 1466

JO - IEEE Transactions on Knowledge and Data Engineering

JF - IEEE Transactions on Knowledge and Data Engineering

IS - 11

ER -

Some effective techniques for naive bayes text classification

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this