<s>
In	O
statistical	O
analysis	O
of	O
binary	B-General_Concept
classification	I-General_Concept
,	O
the	O
F-score	B-General_Concept
or	O
F-measure	B-General_Concept
is	O
a	O
measure	O
of	O
a	O
test	O
's	O
accuracy	O
.	O
</s>
<s>
Precision	O
is	O
also	O
known	O
as	O
positive	O
predictive	O
value	O
,	O
and	O
recall	O
is	O
also	O
known	O
as	O
sensitivity	O
in	O
diagnostic	O
binary	B-General_Concept
classification	I-General_Concept
.	O
</s>
<s>
The	O
F1	B-General_Concept
score	I-General_Concept
is	O
the	O
harmonic	O
mean	O
of	O
the	O
precision	O
and	O
recall	O
.	O
</s>
<s>
The	O
highest	O
possible	O
value	O
of	O
an	O
F-score	B-General_Concept
is	O
1.0	O
,	O
indicating	O
perfect	O
precision	O
and	O
recall	O
,	O
and	O
the	O
lowest	O
possible	O
value	O
is	O
0	O
,	O
if	O
either	O
precision	O
or	O
recall	O
are	O
zero	O
.	O
</s>
<s>
The	O
name	O
F-measure	B-General_Concept
is	O
believed	O
to	O
be	O
named	O
after	O
a	O
different	O
F	O
function	O
in	O
Van	O
Rijsbergen	O
's	O
book	O
,	O
when	O
introduced	O
to	O
the	O
Fourth	O
Message	B-General_Concept
Understanding	I-General_Concept
Conference	I-General_Concept
(	O
MUC-4	O
,	O
1992	O
)	O
.	O
</s>
<s>
The	O
traditional	O
F-measure	B-General_Concept
or	O
balanced	O
F-score	B-General_Concept
(	O
F1	B-General_Concept
score	I-General_Concept
)	O
is	O
the	O
harmonic	O
mean	O
of	O
precision	O
and	O
recall	O
:	O
</s>
<s>
A	O
more	O
general	O
F	B-General_Concept
score	I-General_Concept
,	O
,	O
that	O
uses	O
a	O
positive	O
real	O
factor	O
,	O
where	O
is	O
chosen	O
such	O
that	O
recall	O
is	O
considered	O
times	O
as	O
important	O
as	O
precision	O
,	O
is	O
:	O
</s>
<s>
The	O
F-measure	B-General_Concept
was	O
derived	O
so	O
that	O
"	O
measures	O
the	O
effectiveness	O
of	O
retrieval	O
with	O
respect	O
to	O
a	O
user	O
who	O
attaches	O
times	O
as	O
much	O
importance	O
to	O
recall	O
as	O
precision	O
"	O
.	O
</s>
<s>
This	O
is	O
related	O
to	O
the	O
field	O
of	O
binary	B-General_Concept
classification	I-General_Concept
where	O
recall	O
is	O
often	O
termed	O
"	O
sensitivity	O
"	O
.	O
</s>
<s>
The	O
F-score	B-General_Concept
is	O
often	O
used	O
in	O
the	O
field	O
of	O
information	B-Library
retrieval	I-Library
for	O
measuring	O
search	B-Application
,	O
document	B-Algorithm
classification	I-Algorithm
,	O
and	O
query	B-Library
classification	O
performance	O
.	O
</s>
<s>
Earlier	O
works	O
focused	O
primarily	O
on	O
the	O
F1	B-General_Concept
score	I-General_Concept
,	O
but	O
with	O
the	O
proliferation	O
of	O
large	O
scale	O
search	B-Application
engines	I-Application
,	O
performance	O
goals	O
changed	O
to	O
place	O
more	O
emphasis	O
on	O
either	O
precision	O
or	O
recall	O
and	O
so	O
is	O
seen	O
in	O
wide	O
application	O
.	O
</s>
<s>
The	O
F-score	B-General_Concept
is	O
also	O
used	O
in	O
machine	O
learning	O
.	O
</s>
<s>
However	O
,	O
the	O
F-measures	B-General_Concept
do	O
not	O
take	O
true	O
negatives	O
into	O
account	O
,	O
hence	O
measures	O
such	O
as	O
the	O
Matthews	B-General_Concept
correlation	I-General_Concept
coefficient	I-General_Concept
,	O
Informedness	B-General_Concept
or	O
Cohen	B-General_Concept
's	I-General_Concept
kappa	I-General_Concept
may	O
be	O
preferred	O
to	O
assess	O
the	O
performance	O
of	O
a	O
binary	B-General_Concept
classifier	I-General_Concept
.	O
</s>
<s>
The	O
F-score	B-General_Concept
has	O
been	O
widely	O
used	O
in	O
the	O
natural	O
language	O
processing	O
literature	O
,	O
such	O
as	O
in	O
the	O
evaluation	O
of	O
named	B-General_Concept
entity	I-General_Concept
recognition	I-General_Concept
and	O
word	O
segmentation	O
.	O
</s>
<s>
The	O
F1	B-General_Concept
score	I-General_Concept
is	O
the	O
Dice	O
coefficient	O
of	O
the	O
set	O
of	O
retrieved	O
items	O
and	O
the	O
set	O
of	O
relevant	O
items	O
.	O
</s>
<s>
David	O
Hand	O
and	O
others	O
criticize	O
the	O
widespread	O
use	O
of	O
the	O
F1	B-General_Concept
score	I-General_Concept
since	O
it	O
gives	O
equal	O
importance	O
to	O
precision	O
and	O
recall	O
.	O
</s>
<s>
According	O
to	O
Davide	O
Chicco	O
and	O
Giuseppe	O
Jurman	O
,	O
the	O
F1	B-General_Concept
score	I-General_Concept
is	O
less	O
truthful	O
and	O
informative	O
than	O
the	O
Matthews	B-General_Concept
correlation	I-General_Concept
coefficient	I-General_Concept
(	O
MCC	O
)	O
in	O
binary	O
evaluation	O
classification	O
.	O
</s>
<s>
David	O
Powers	O
has	O
pointed	O
out	O
that	O
F1	O
ignores	O
the	O
True	O
Negatives	O
and	O
thus	O
is	O
misleading	O
for	O
unbalanced	O
classes	O
,	O
while	O
kappa	O
and	O
correlation	O
measures	O
are	O
symmetric	O
and	O
assess	O
both	O
directions	O
of	O
predictability	O
-	O
the	O
classifier	O
predicting	O
the	O
true	O
class	O
and	O
the	O
true	O
class	O
predicting	O
the	O
classifier	O
prediction	O
,	O
proposing	O
separate	O
multiclass	O
measures	O
Informedness	B-General_Concept
and	O
Markedness	O
for	O
the	O
two	O
directions	O
,	O
noting	O
that	O
their	O
geometric	O
mean	O
is	O
correlation	O
.	O
</s>
<s>
This	O
criticism	O
is	O
met	O
by	O
the	O
P4	B-General_Concept
metric	I-General_Concept
definition	O
,	O
which	O
is	O
sometimes	O
indicated	O
as	O
a	O
symmetrical	O
extension	O
of	O
F1	O
.	O
</s>
<s>
While	O
the	O
F-measure	B-General_Concept
is	O
the	O
harmonic	O
mean	O
of	O
recall	O
and	O
precision	O
,	O
the	O
Fowlkes	O
–	O
Mallows	O
index	O
is	O
their	O
geometric	O
mean	O
.	O
</s>
<s>
The	O
F-score	B-General_Concept
is	O
also	O
used	O
for	O
evaluating	O
classification	O
problems	O
with	O
more	O
than	O
two	O
classes	O
(	O
Multiclass	B-General_Concept
classification	I-General_Concept
)	O
.	O
</s>
<s>
For	O
macro-averaging	O
,	O
two	O
different	O
formulas	O
have	O
been	O
used	O
by	O
applicants	O
:	O
the	O
F-score	B-General_Concept
of	O
(	O
arithmetic	O
)	O
class-wise	O
precision	O
and	O
recall	O
means	O
or	O
the	O
arithmetic	O
mean	O
of	O
class-wise	O
F-scores	B-General_Concept
,	O
where	O
the	O
latter	O
exhibits	O
more	O
desirable	O
properties	O
.	O
</s>
