This research explores the extent to which one can abstract away the major components of decision tree building to facilitate the development of new decision tree algorithms by easily embedding them in a library that supports all tools required for algorithm evaluation (for example, I/O management, classification, visualizations, etc.)
The software (DTToolkit) has been used as a testbed for a programming laboratory with junior undergraduates and, from time to time, is actively maintained and undergoing quality assurance.
The software is available upon demand for academic use (but all documentation is in Greek).
The research is described in the following papers:
D. Kalles and A. Papagelis. “Managing the Decision Tree Life-Cycle with Components”, International Journal of Information and Communications Technology Education, Vol. 2, Iss. 3, pp. 1-13, 2006.
S. Christodoulou, K. Hantzara, D. Kalles, and A. Papagelis. “Building Decision Trees with Components”, Panhellenic Conference on Artificial Intelligence, Samos, Greece, May 2004.
N. Drossos, A. Papagelis and D. Kalles. “Decision Tree Toolkit: A Component-based Library of Decision Tree Algorithms”, 4th European Conference on Principles and Practice of Knowledge Discovery in Databases, Lyon, France, 2000.
The teaching experience is described in the educational technology section.