<s>
Text	B-General_Concept
Nailing	I-General_Concept
(	O
TN	O
)	O
is	O
an	O
information	B-General_Concept
extraction	I-General_Concept
method	O
of	O
semi-automatically	O
extracting	O
structured	O
information	O
from	O
unstructured	O
documents	O
.	O
</s>
<s>
The	O
identified	O
expressions	O
can	O
be	O
used	O
then	O
to	O
enhance	O
computational	O
methods	O
that	O
rely	O
on	O
text	O
(	O
e.g.	O
,	O
Regular	B-Language
expression	I-Language
)	O
as	O
well	O
as	O
advanced	O
natural	B-Language
language	I-Language
processing	I-Language
(	O
NLP	B-Language
)	O
techniques	O
.	O
</s>
<s>
In	O
traditional	O
machine	O
learning	O
approaches	O
for	O
text	B-Algorithm
classification	I-Algorithm
,	O
a	O
human	O
expert	O
is	O
required	O
to	O
label	O
phrases	O
or	O
entire	O
notes	O
,	O
and	O
then	O
a	O
supervised	B-General_Concept
learning	I-General_Concept
algorithm	O
attempts	O
to	O
generalize	O
the	O
associations	O
and	O
apply	O
them	O
to	O
new	O
data	O
.	O
</s>
<s>
A	O
comprehensive	O
review	O
regarding	O
extracting	O
information	O
from	O
textual	O
documents	O
in	O
the	O
electronic	B-Application
health	I-Application
record	I-Application
is	O
available	O
.	O
</s>
<s>
A	O
sample	O
code	O
for	O
extracting	O
smoking	O
status	O
from	O
narrative	O
notes	O
using	O
"	O
nailed	O
expressions	O
"	O
is	O
available	O
in	O
GitHub	B-Application
.	O
</s>
<s>
In	O
his	O
viewpoint	O
published	O
in	O
June	O
2018	O
concerning	O
slow	O
adoption	O
of	O
data-driven	O
findings	O
in	O
medicine	O
,	O
Uri	O
Kartoun	O
,	O
co-creator	O
of	O
Text	B-General_Concept
Nailing	I-General_Concept
states	O
that	O
"	O
Text	O
...	O
Nailing	O
raised	O
skepticism	O
in	O
reviewers	O
of	O
medical	O
informatics	O
journals	O
who	O
claimed	O
that	O
it	O
relies	O
on	O
simple	O
tricks	O
to	O
simplify	O
the	O
text	O
,	O
and	O
leans	O
heavily	O
on	O
human	O
annotation	O
.	O
</s>
<s>
In	O
this	O
case	O
the	O
features	O
the	O
experts	O
create	O
are	O
equivalent	O
to	O
regular	B-Language
expressions	I-Language
.	O
</s>
<s>
Removing	O
non-alphabetical	O
characters	O
and	O
matching	O
on	O
"	O
smokesppd	O
"	O
is	O
equal	O
to	O
the	O
regular	B-Language
expression	I-Language
/smokes[^a-zA-Z]*ppd/	O
.	O
</s>
<s>
Using	O
regular	B-Language
expressions	I-Language
as	O
features	O
for	O
text	B-Algorithm
classification	I-Algorithm
is	O
not	O
novel	O
.	O
</s>
<s>
The	O
authors	O
note	O
that	O
using	O
support	B-Algorithm
vector	I-Algorithm
machines	I-Algorithm
(	O
SVM	B-Algorithm
)	O
and	O
hundreds	O
of	O
documents	O
give	O
inferior	O
performance	O
,	O
but	O
does	O
not	O
specify	O
which	O
features	O
or	O
documents	O
the	O
SVM	B-Algorithm
was	O
trained/tested	O
on	O
.	O
</s>
