Gene clustering: 2001:
2
Finding Rules
§
Words
of cluster
characterise
its
instances
§
Construct rules with
•
“necessary” words
•
“sufficient” words
•
“supplementary” words
§
Construct “ordinary” discriminant rules.
if
E.coli
&
(intergenic
or
(hypothetical
&
region))
then
UP
§
Not all words necessary to
discriminate
the
cluster
•
eg
, if all instances in dataset have word
“
protein"
then
“
protein"
is in every cluster
but
“
protein"
is not useful
to discriminate
UP
if
•
E.coli
•
intergenic, hypothetical, region
•
protein, in, to
