DYNAMIC CLUSTERING FOR TIME INCREMENTAL DATA

The aim of this article is to propose a model for on-line data cluster ing when a new subset of data accumulates after an interval of time. T he new data may be absorbed in the old clusters or form new clusters o r appear as stray data. The absorbed data may cause the clusters to gr ow so that two grown clusters may merge to form a single cluster. On t he other hand, a large number of absorbed data may change the density profile of a cluster so that it should be split into two or more clust ers. Procedures to compute these situations are proposed.