Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization

Yaxin Bi, David A. Bell, Hui Wang, Gongde Guo, Kieran Greer

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

In this paper, we present an investigation into the combination of four different classification methods for text categorization using Dempster’s rule of combination. These methods include the Support Vector Machine, kNN (nearest neighbours), kNN model-based approach (kNNM), and Rocchio methods. We first present an approach for effectively combining the different classification methods. We then apply these methods to a benchmark data collection of 20-newsgroup, individually and in combination. Our experimental results show that the performance of the best combination of the different classifiers on the 10 groups of the benchmark data can achieve 91.07% classification accuracy, which is 2.68% better than that of the best individual method, SVM, on average.
LanguageEnglish
Title of host publicationModeling Decisions for Artificial Intelligence Lecture Notes in Computer Science
Pages127-138
Publication statusPublished - 2004

Fingerprint

Classifiers
Support vector machines

Cite this

Bi, Y., Bell, D. A., Wang, H., Guo, G., & Greer, K. (2004). Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization. In Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science (pp. 127-138)
Bi, Yaxin ; Bell, David A. ; Wang, Hui ; Guo, Gongde ; Greer, Kieran. / Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization. Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science. 2004. pp. 127-138
@inbook{1f0440f3c08f48f8bec142ec3b5849a4,
title = "Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization",
abstract = "In this paper, we present an investigation into the combination of four different classification methods for text categorization using Dempster’s rule of combination. These methods include the Support Vector Machine, kNN (nearest neighbours), kNN model-based approach (kNNM), and Rocchio methods. We first present an approach for effectively combining the different classification methods. We then apply these methods to a benchmark data collection of 20-newsgroup, individually and in combination. Our experimental results show that the performance of the best combination of the different classifiers on the 10 groups of the benchmark data can achieve 91.07{\%} classification accuracy, which is 2.68{\%} better than that of the best individual method, SVM, on average.",
author = "Yaxin Bi and Bell, {David A.} and Hui Wang and Gongde Guo and Kieran Greer",
year = "2004",
language = "English",
isbn = "978-3-540-22555-3",
pages = "127--138",
booktitle = "Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science",

}

Bi, Y, Bell, DA, Wang, H, Guo, G & Greer, K 2004, Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization. in Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science. pp. 127-138.

Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization. / Bi, Yaxin; Bell, David A.; Wang, Hui; Guo, Gongde; Greer, Kieran.

Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science. 2004. p. 127-138.

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization

AU - Bi, Yaxin

AU - Bell, David A.

AU - Wang, Hui

AU - Guo, Gongde

AU - Greer, Kieran

PY - 2004

Y1 - 2004

N2 - In this paper, we present an investigation into the combination of four different classification methods for text categorization using Dempster’s rule of combination. These methods include the Support Vector Machine, kNN (nearest neighbours), kNN model-based approach (kNNM), and Rocchio methods. We first present an approach for effectively combining the different classification methods. We then apply these methods to a benchmark data collection of 20-newsgroup, individually and in combination. Our experimental results show that the performance of the best combination of the different classifiers on the 10 groups of the benchmark data can achieve 91.07% classification accuracy, which is 2.68% better than that of the best individual method, SVM, on average.

AB - In this paper, we present an investigation into the combination of four different classification methods for text categorization using Dempster’s rule of combination. These methods include the Support Vector Machine, kNN (nearest neighbours), kNN model-based approach (kNNM), and Rocchio methods. We first present an approach for effectively combining the different classification methods. We then apply these methods to a benchmark data collection of 20-newsgroup, individually and in combination. Our experimental results show that the performance of the best combination of the different classifiers on the 10 groups of the benchmark data can achieve 91.07% classification accuracy, which is 2.68% better than that of the best individual method, SVM, on average.

M3 - Chapter

SN - 978-3-540-22555-3

SP - 127

EP - 138

BT - Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science

ER -

Bi Y, Bell DA, Wang H, Guo G, Greer K. Combining Multiple Classifiers Using Dempster's Rule of Combination for Text Categorization. In Modeling Decisions for Artificial Intelligence Lecture Notes in Computer Science. 2004. p. 127-138