Giga-Mining

Corinna Cortes and Daryl Pregibon

We describe an industrial-strength data mining application in telecommunications. The application requires building a short (7 byte) profile for all telephone numbers seen on a large telecom network. By large, we mean very large: we maintain approximately 350 million profiles. In addition, the procedure for updating these profiles is based on processing approximately 275 million call records per day. We discuss the motivation for massive tracking and fully describe the definition and the computation of one of the more interesting bytes in the profile.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.