<s>
Coupled	B-General_Concept
Pattern	I-General_Concept
Learner	I-General_Concept
(	O
CPL	O
)	O
is	O
a	O
machine	O
learning	O
algorithm	O
which	O
couples	O
the	O
semi-supervised	B-General_Concept
learning	I-General_Concept
of	O
categories	O
and	O
relations	O
to	O
forestall	O
the	O
problem	O
of	O
semantic	O
drift	O
associated	O
with	O
boot-strap	O
learning	O
methods	O
.	O
</s>
<s>
Semi-supervised	B-General_Concept
learning	I-General_Concept
approaches	O
using	O
a	O
small	O
number	O
of	O
labeled	O
examples	O
with	O
many	O
unlabeled	O
examples	O
are	O
usually	O
unreliable	O
as	O
they	O
produce	O
an	O
internally	O
consistent	O
,	O
but	O
incorrect	O
set	O
of	O
extractions	O
.	O
</s>
<s>
CPL	O
solves	O
this	O
problem	O
by	O
simultaneously	O
learning	O
classifiers	O
for	O
many	O
different	O
categories	O
and	O
relations	O
in	O
the	O
presence	O
of	O
an	O
ontology	B-Language
defining	O
constraints	O
that	O
couple	O
the	O
training	O
of	O
these	O
classifiers	O
.	O
</s>
<s>
CPL	O
is	O
an	O
approach	O
to	O
semi-supervised	B-General_Concept
learning	I-General_Concept
that	O
yields	O
more	O
accurate	O
results	O
by	O
coupling	O
the	O
training	O
of	O
many	O
information	O
extractors	O
.	O
</s>
<s>
CPL	O
primarily	O
relies	O
on	O
the	O
notion	O
of	O
coupling	O
the	O
learning	O
of	O
multiple	O
functions	O
so	O
as	O
to	O
constrain	O
the	O
semi-supervised	B-General_Concept
learning	I-General_Concept
problem	O
.	O
</s>
<s>
Each	O
predicate	O
P	O
in	O
the	O
ontology	B-Language
has	O
a	O
list	O
of	O
other	O
same-arity	O
predicates	O
with	O
which	O
P	O
is	O
mutually	O
exclusive	O
.	O
</s>
<s>
A	O
large	O
corpus	O
of	O
Part-Of-Speech	O
tagged	O
sentences	O
and	O
an	O
initial	O
ontology	B-Language
with	O
predefined	O
categories	O
,	O
relations	O
,	O
mutually	O
exclusive	O
relationships	O
between	O
same-arity	O
predicates	O
,	O
subset	O
relationships	O
between	O
some	O
categories	O
,	O
seed	O
instances	O
for	O
all	O
predicates	O
,	O
and	O
seed	O
patterns	O
for	O
the	O
categories	O
.	O
</s>
