Abstract
Text categorization is an important application domain of multi-label classification where each document can simultaneously belong to more than one class. The most common approach is to address the problem of multi-label examples by inducing a separate binary classifier for each class, and then use these classifiers in parallel. What the information-retrieval community has all but ignored, however, is that such classifiers are almost always induced from highly imbalanced training sets. The study reported in this paper shows how taking this aspect into consideration with a majority-class undersampling we used here can indeed improve classification performance as measured by criteria common in text categorization: macro/micro precision, recall, and F 1. We also show how a slight modification of an older undersampling technique helps further improve the results.
| Original language | English |
|---|---|
| Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
| Pages | 40-52 |
| Number of pages | 13 |
| Volume | 5669 LNAI |
| DOIs | |
| State | Published - Sep 30 2010 |
| Event | 13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009 - Bangkok, Thailand Duration: Apr 27 2009 → Apr 30 2009 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 5669 LNAI |
| ISSN (Print) | 03029743 |
| ISSN (Electronic) | 16113349 |
Other
| Other | 13th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2009 |
|---|---|
| Country | Thailand |
| City | Bangkok |
| Period | 4/27/09 → 4/30/09 |
Fingerprint
ASJC Scopus subject areas
- Computer Science(all)
- Theoretical Computer Science
Cite this
Undersampling approach for imbalanced training sets and induction from multi-label text-categorization domains. / Dendamrongvit, Sareewan; Kubat, Miroslav.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 5669 LNAI 2010. p. 40-52 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5669 LNAI).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Undersampling approach for imbalanced training sets and induction from multi-label text-categorization domains
AU - Dendamrongvit, Sareewan
AU - Kubat, Miroslav
PY - 2010/9/30
Y1 - 2010/9/30
N2 - Text categorization is an important application domain of multi-label classification where each document can simultaneously belong to more than one class. The most common approach is to address the problem of multi-label examples by inducing a separate binary classifier for each class, and then use these classifiers in parallel. What the information-retrieval community has all but ignored, however, is that such classifiers are almost always induced from highly imbalanced training sets. The study reported in this paper shows how taking this aspect into consideration with a majority-class undersampling we used here can indeed improve classification performance as measured by criteria common in text categorization: macro/micro precision, recall, and F 1. We also show how a slight modification of an older undersampling technique helps further improve the results.
AB - Text categorization is an important application domain of multi-label classification where each document can simultaneously belong to more than one class. The most common approach is to address the problem of multi-label examples by inducing a separate binary classifier for each class, and then use these classifiers in parallel. What the information-retrieval community has all but ignored, however, is that such classifiers are almost always induced from highly imbalanced training sets. The study reported in this paper shows how taking this aspect into consideration with a majority-class undersampling we used here can indeed improve classification performance as measured by criteria common in text categorization: macro/micro precision, recall, and F 1. We also show how a slight modification of an older undersampling technique helps further improve the results.
UR - http://www.scopus.com/inward/record.url?scp=77957042586&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77957042586&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-14640-4_4
DO - 10.1007/978-3-642-14640-4_4
M3 - Conference contribution
AN - SCOPUS:77957042586
SN - 3642146392
SN - 9783642146398
VL - 5669 LNAI
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 40
EP - 52
BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ER -