Universitat Oberta de Catalunya
Computing frequent itemsets is a central data mining task, both in the static and the streaming scenarios. Important research effort has produced a substantial number of methods for the streaming case, and the problem is relatively well understood now. The authors noticed, however, that there are almost no public, easy-to-use implementations of the methods described in the literature, a situation that effectively prevents their application in practice and conditions further research. This paper is to describe a robust, efficient, usable, and extensible implementation for mining frequent closed itemsets over data streams, working over the MOA framework.