|
About
NeuroCOLT
Papers
Archive
Books
info@neurocolt.org
|
NeuroCOLT
Technical Report NC-TR-01-091
2001-091
Mining
Unbounded Episodes from Sequential Data
J Baixeries, G Casas-Garriga,
JL Balcazar
ABSTRACT
The discovery of frequent sequential patterns in an ordered collection
of data is an important data mining issue. In this problem, we are
given a sequence of events, where each event has an associated time
of occurrence. One basic goal in analyzing such sequences is to find
frequent interesting episodes, i.e, collections of events occurring
frequently together in the input sequence. We present a new definition
of interesting ness for episodes, along with algorithms to efficiently
discover them. Most current work decides the interestingness of an
episode from a fixed user-specified window width; we introduce a more
intuitive definition that allows, in turn, interesting episodes to
grow during the mining without any user-specified help. Experimental
results confirm that our approach results useful and advantageous.
Download
Postscript
|