Enhancing recommendations through a data mining algorithm

Abstract

This paper reports on the development of a new data mining algorithm that formulates purposeful association rules out of the transactions' database of a transportation management system,. The proposed algorithm is generic and capable to construct such rules by creating a large set of related items. The constructed rules can be used by the system's recommender module, which is responsible for providing recommendations to the associated users. The recommendation process takes into account the constructed rules and techniques that derive from the area of collaborative filtering. Our approach enables users to receive high quality recommendations for their upcoming transactions. © 2008 Springer-Verlag Berlin Heidelberg.