Thomas Moore, Larry Hall and Kevin Bowyer

Department of Computer Science and Engineering

University of South Florida

Tampa, Fl 33620

tmoore4 | hall | kwb @csee.usf.edu

fax: 813.974.5456

contact for correspondence: tmoore4@csee.usf.edu

Parallel classifiers are important in the development of data mining tools for massive data sets. The ScalParC parallel decision tree system is a well-known example. We identify a problem in the ScalParC implementation and discuss the effects of correcting this problem on the scalability of the approach.