The analysis results show that elder self-care behavior can be
classified by an interest-based representation scheme into three
distinct cluster types: health-dominate type, entertainment-dominate
type and general-dominate type. Each type displays different
self-care use cycles, times, function numbers and needs. However,
six distinct cluster types can be identified by sequence-based clustering
from different use sequence. Each type provides detailed
information about self-care use cycle, time, function numbers
and characterizations. This research shows that the use of sequence-
based clustering in web usage mining effectively finds
meaning groups that share common interests and behaviors and
effectively extracts knowledge needed to understand the motivation
for using elder self-care. The analysis results can be used by
experts in medicine, public health, nursing and psychology to further
research and to assist in policy-making in the health care domain.
Future research will apply the sequence-based clustering
results for the ComCare project to improving personalized elder
self-care services.