Deakin home > Deakin University Library > Deakin Research Online > Reducing performance bias for unbalanced text mining

Reducing performance bias for unbalanced text mining

Zhuang, Ling and Dai, Honghua 2006, Reducing performance bias for unbalanced text mining, in ICDM Workshops 2006 proceedings : 18 December, 2006, Hong Kong, China, IEEE Computer Society, Los Alamitos, Calif., pp. 770-774.

Document type: Conference Paper
Collection: School of Engineering and Information Technology
Attached Files (Some files may be inaccessible until you login with your Deakin Research Online credentials)
Name Description MIMEType Size Downloads

Title Reducing performance bias for unbalanced text mining
Author(s) Zhuang, Ling
Dai, Honghua
Conference name Sixth IEEE International Conference on Data Mining - Workshops (ICDMW'06)
Conference location Hong Kong, China
Conference dates 18-22 December 2006
Title of proceedings ICDM Workshops 2006 proceedings : 18 December, 2006, Hong Kong, China
Editor(s) Tsumoto, Shusaku
Clifton, Christopher W.
Zhong, Ning
Wu, Xindong
Liu, Jiming
Wah, Benjamin W.
Cheung, Yiu-Ming
Publication date 2006
Conference series IEEE International Conference on Data Mining Workshops
Start page 770
End page 774
Publisher IEEE Computer Society
Place of publication Los Alamitos, Calif.
Summary In text categorization applications, class imbalance, which refers to an uneven data distribution where one class is represented by far more less instances than the others, is a commonly encountered problem. In such a situation, conventional classifiers tend to have a strong performance bias, which results in high accuracy rate on the majority class but very low rate on the minorities. An extreme strategy for unbalanced, learning is to discard the majority instances and apply one-class classification to the minority class. However, this could easily cause another type of bias, which increases the accuracy rate on minorities by sacrificing the majorities. This paper aims to investigate approaches that reduce these two types of performance bias and improve the reliability of discovered classification rules. Experimental results show that the inexact field learning method and parameter optimized one-class classifiers achieve more balanced performance than the standard approaches.
Language eng
Field of Research 080105 Expert Systems
HERDC Research category E1 Full written paper - refereed
Copyright notice ©2006, IEEE
Persistent URL http://hdl.handle.net/10536/DRO/DU:30006156
Connect to link resolver
 
Unless expressly stated otherwise, the copyright for items in Deakin Research Online is owned by the author, with all rights reserved.
Versions
Version Filter Type
Access Statistics: 452 Abstract Views, 0 File Downloads  -  Detailed Statistics
Created: Mon, 07 Jul 2008, 09:59:14 EST