<s>
Information	B-Algorithm
Harvesting	I-Algorithm
(	O
IH	O
)	O
was	O
an	O
early	O
data	B-Application
mining	I-Application
product	O
from	O
the	O
1990s	O
.	O
</s>
<s>
It	O
was	O
invented	O
by	O
Ralphe	O
Wiggins	O
and	O
produced	O
by	O
the	O
Ryan	O
Corp	O
,	O
later	O
Information	B-Algorithm
Harvesting	I-Algorithm
Inc.	O
,	O
of	O
Cambridge	O
,	O
Massachusetts	O
.	O
</s>
<s>
Wiggins	O
had	O
a	O
background	O
in	O
genetic	B-Algorithm
algorithms	I-Algorithm
and	O
fuzzy	O
logic	O
.	O
</s>
<s>
It	O
included	O
strategies	O
for	O
checking	O
if	O
overfitting	B-Error_Name
took	O
place	O
and	O
,	O
if	O
so	O
,	O
correcting	O
for	O
it	O
.	O
</s>
<s>
Because	O
of	O
its	O
strategies	O
for	O
correcting	O
for	O
overfitting	B-Error_Name
by	O
considering	O
more	O
data	O
,	O
and	O
refining	O
the	O
rules	O
based	O
on	O
that	O
data	O
,	O
IH	O
might	O
also	O
be	O
considered	O
to	O
be	O
a	O
form	O
of	O
machine	O
learning	O
.	O
</s>
<s>
The	O
advantage	O
of	O
IH	O
,	O
as	O
compared	O
with	O
other	O
data	B-Application
mining	I-Application
products	O
of	O
its	O
time	O
and	O
even	O
later	O
,	O
was	O
that	O
it	O
provided	O
a	O
mechanism	O
for	O
finding	O
multiple	O
rules	O
that	O
would	O
classify	O
the	O
data	O
and	O
determining	O
,	O
according	O
to	O
set	O
criteria	O
,	O
the	O
best	O
rules	O
to	O
use	O
.	O
</s>
