Temporal Data Mining

92    ◾    Temporal Data Mining

3.2.2.1 The COBWEB Algorithm

is algorithm is described in [Fis87] and is an incremental, hierarchi-

cal, conceptual clustering algorithm, in which each node in the hierar-

chy represents a concept and describes probabilistically the objects that

belong to this concept. COBWEB is incremental, in the sense that objects

are added incrementally to the hierarchical tree in a top-down approach

looking for the best match. While descending the nodes, an object can be

assigned to a cluster (using the incorporate operator), a new cluster can be

created (with the create operator), a cluster can be divided (with the split

operator), or two clusters can be merged in one (with the merge operator).

Which operation will be