Cube Based Incremental Data Mining

This work presents a cube-based approach forincremental data mining that operates on the datarather than on the data mining algorithms. The ideawas to build a compressed replica of the full-blowndatabase by representing the database by means ofmulti-dimensional cubes and then applying theoriginal data mining algorithms on the cube-baseddata. This way the storage requirement toaccommodate the database is not affected by the newdata. Yet the fact that we used the original datamining algorithms on the cube-based data makes ourincremental data mining approach very general as itcan be applied to all types of data mining models.