In Web usage mining, clustering finds groups that share common
properties and behavior by analyzing the data collected in
web servers. Given the transformation of elder self-care access sessions
into a multi-dimensional space as interest-based representation
vectors or sequence-based representation matrices of
functions, a clustering algorithm was applied to the derived elder
self-care access sessions. Since access sessions are the images of
activities by elders, representative elder self-care patterns can be
obtained by clustering. These patterns also facilitate profiling of elder
users of the ComCare service. This section describes how session
clustering is performed and how cluster number is
determined.