Research Article

A Dynamic Ensemble Framework for Mining Textual Streams with Class Imbalance

Table 1

The properties of textual streams.

The number of instancesImbalanced radioSize of attributeSize of each chunk
SpamLegitimateTotal

Spam Assassin2,3876,9379,3241 : 3500300
Enron Email17,15716,54533,7021 : 11,545
Spam-Enron25,00075,000100,0001 : 32,044500
Spam11,187 6,9378,1001 : 6500300
Spam22806,9207,2001 : 25500300
Reuters-Spam9569,04410,0001 : 1019,433500