This work presents a cube-based approach for
incremental data mining that operates on the data,
rather than on the data mining algorithms. The idea
was to build a compressed replica of the full-blown
database by representing the database by means of
multi-dimensional cubes, and then applying the
original data mining algorithms on the cube-based
data. This way, the storage requirement to
accommodate the database is not affected by the new
data. Yet, the fact that we used the original data
mining algorithms on the cube-based data makes our
incremental data mining approach very general as it
can be applied to all types of data mining models.
incremental data mining that operates on the data,
rather than on the data mining algorithms. The idea
was to build a compressed replica of the full-blown
database by representing the database by means of
multi-dimensional cubes, and then applying the
original data mining algorithms on the cube-based
data. This way, the storage requirement to
accommodate the database is not affected by the new
data. Yet, the fact that we used the original data
mining algorithms on the cube-based data makes our
incremental data mining approach very general as it
can be applied to all types of data mining models.