Survey on data stream classification technologies
QU Ya-hui · Jisuanji yingyong yanjiu · 2009
The high speed,continuous and virtual infinite,and time-changing inherent features of streaming data invalidate the traditional technologies of data analysis and data mining,or motivate the need for evolving themselves.This paper put emphasis on data stream classification,proposed a few critical issues accompanying with data stream classification,surveyed typical classification approaches.Furthermore,presented a new framework exploiting active learning and semi-supervised learning to attack the disadvantages of current technologies.