<s>
Collaborative	B-General_Concept
tagging	I-General_Concept
,	O
also	O
known	O
as	O
social	B-General_Concept
tagging	I-General_Concept
or	O
folksonomy	B-General_Concept
,	O
allows	O
users	O
to	O
apply	O
public	O
tags	B-General_Concept
to	O
online	O
items	O
,	O
typically	O
to	O
make	O
those	O
items	O
easier	O
for	O
themselves	O
or	O
others	O
to	O
find	O
later	O
.	O
</s>
<s>
The	O
notion	O
is	O
that	O
given	O
that	O
social	O
tags	B-General_Concept
are	O
labels	O
users	O
create	O
to	O
represent	O
topics	O
extracted	O
from	O
online	O
documents	O
,	O
the	O
interpretation	O
of	O
these	O
tags	B-General_Concept
should	O
allow	O
other	O
users	O
to	O
predict	O
the	O
contents	O
of	O
different	O
documents	O
efficiently	O
.	O
</s>
<s>
Social	O
tags	B-General_Concept
are	O
arguably	O
more	O
important	O
in	O
exploratory	O
search	O
,	O
in	O
which	O
the	O
users	O
may	O
engage	O
in	O
iterative	O
cycles	O
of	O
goal	O
refinement	O
and	O
exploration	O
of	O
new	O
information	O
(	O
as	O
opposed	O
to	O
simple	O
fact-retrievals	O
)	O
,	O
and	O
interpretation	O
of	O
information	O
contents	O
by	O
others	O
will	O
provide	O
useful	O
cues	O
for	O
people	O
to	O
discover	O
topics	O
that	O
are	O
relevant	O
.	O
</s>
<s>
One	O
significant	O
challenge	O
that	O
arises	O
in	O
social	B-General_Concept
tagging	I-General_Concept
systems	O
is	O
the	O
rapid	O
increase	O
in	O
the	O
number	O
and	O
diversity	O
of	O
tags	B-General_Concept
.	O
</s>
<s>
As	O
opposed	O
to	O
structured	O
annotation	O
systems	O
,	O
tags	B-General_Concept
provide	O
users	O
an	O
unstructured	O
,	O
open-ended	O
mechanism	O
to	O
annotate	O
and	O
organize	O
web	O
content	O
.	O
</s>
<s>
As	O
users	O
are	O
free	O
to	O
create	O
any	O
tag	B-General_Concept
to	O
describe	O
any	O
resource	O
,	O
it	O
leads	O
to	O
what	O
is	O
referred	O
to	O
as	O
the	O
vocabulary	O
problem	O
.	O
</s>
<s>
Because	O
users	O
may	O
use	O
different	O
words	O
to	O
describe	O
the	O
same	O
document	O
or	O
extract	O
different	O
topics	O
from	O
the	O
same	O
document	O
based	O
on	O
their	O
own	O
background	O
knowledge	O
,	O
the	O
lack	O
of	O
any	O
top-down	O
mediation	O
may	O
lead	O
to	O
an	O
increase	O
in	O
the	O
use	O
of	O
incoherent	O
tags	B-General_Concept
to	O
represent	O
the	O
information	O
resources	O
in	O
the	O
system	O
.	O
</s>
<s>
In	O
other	O
words	O
,	O
the	O
lack	O
of	O
structure	O
inherent	O
in	O
social	O
tags	B-General_Concept
may	O
hinder	O
their	O
potential	O
as	O
navigational	O
cues	O
for	O
searchers	O
because	O
the	O
diversities	O
of	O
users	O
and	O
their	O
motivation	O
may	O
lead	O
to	O
diminishing	O
tag-topic	O
relations	O
as	O
the	O
system	O
grows	O
.	O
</s>
<s>
However	O
,	O
a	O
number	O
of	O
studies	O
have	O
shown	O
that	O
structures	O
do	O
emerge	O
at	O
the	O
semantic	O
level	O
–	O
indicating	O
that	O
there	O
are	O
cohesive	O
forces	O
driving	O
the	O
emergent	O
structures	O
in	O
a	O
social	B-General_Concept
tagging	I-General_Concept
system	O
.	O
</s>
<s>
Just	O
like	O
any	O
social	O
phenomena	O
,	O
behavioral	O
patterns	O
in	O
social	B-General_Concept
tagging	I-General_Concept
systems	O
can	O
be	O
characterized	O
by	O
either	O
a	O
descriptive	B-General_Concept
or	O
predictive	B-General_Concept
model	I-General_Concept
.	O
</s>
<s>
While	O
descriptive	B-General_Concept
models	O
ask	O
the	O
question	O
of	O
"	O
what	O
"	O
,	O
predictive	B-General_Concept
models	I-General_Concept
go	O
deeper	O
to	O
also	O
ask	O
the	O
question	O
of	O
"	O
why	O
"	O
by	O
attempting	O
to	O
provide	O
explanations	O
for	O
the	O
aggregate	O
behavioral	O
patterns	O
.	O
</s>
<s>
Descriptive	B-General_Concept
models	O
typically	O
are	O
not	O
concerned	O
with	O
explaining	O
the	O
actions	O
of	O
individuals	O
.	O
</s>
<s>
Predictive	B-General_Concept
models	I-General_Concept
,	O
however	O
,	O
attempt	O
to	O
explain	O
aggregate	O
patterns	O
by	O
analyzing	O
how	O
individuals	O
interact	O
and	O
link	O
to	O
each	O
other	O
in	O
ways	O
that	O
bring	O
about	O
similar	O
or	O
different	O
emergent	O
patterns	O
of	O
social	O
behavior	O
.	O
</s>
<s>
In	O
particular	O
,	O
a	O
mechanism-based	O
predictive	B-General_Concept
model	I-General_Concept
assumes	O
a	O
certain	O
set	O
of	O
rules	O
governing	O
how	O
individuals	O
interact	O
with	O
each	O
other	O
,	O
and	O
understand	O
how	O
these	O
interactions	O
could	O
produce	O
aggregate	O
patterns	O
as	O
observed	O
and	O
characterized	O
by	O
descriptive	B-General_Concept
models	O
.	O
</s>
<s>
Predictive	B-General_Concept
models	I-General_Concept
can	O
therefore	O
provide	O
explanations	O
to	O
why	O
different	O
system	O
characteristics	O
may	O
lead	O
to	O
different	O
aggregate	O
patterns	O
,	O
and	O
can	O
therefore	O
potentially	O
provide	O
information	O
on	O
how	O
systems	O
should	O
be	O
designed	O
to	O
achieve	O
different	O
social	O
purposes	O
.	O
</s>
<s>
For	O
most	O
tagging	O
systems	O
,	O
the	O
total	O
number	O
of	O
objects	O
being	O
tagged	O
far	O
exceeds	O
the	O
total	O
number	O
of	O
tags	B-General_Concept
in	O
the	O
collective	O
vocabulary	O
.	O
</s>
<s>
If	O
a	O
single	O
tag	B-General_Concept
in	O
this	O
system	O
is	O
specified	O
,	O
many	O
documents	O
would	O
match	O
,	O
so	O
that	O
using	O
single	O
tags	B-General_Concept
cannot	O
effectively	O
isolate	O
any	O
one	O
document	O
.	O
</s>
<s>
Thus	O
,	O
the	O
focus	O
should	O
be	O
on	O
how	O
well	O
the	O
mapping	O
of	O
tags	B-General_Concept
to	O
documents	O
retains	O
information	O
about	O
the	O
distribution	O
of	O
the	O
documents	O
.	O
</s>
<s>
A	O
2008	O
paper	O
by	O
Ed	O
Chi	O
and	O
Todd	O
Mytkowicz	O
showed	O
that	O
the	O
entropy	O
of	O
documents	O
conditional	O
on	O
tags	B-General_Concept
,	O
H( D|T	O
)	O
,	O
is	O
increasing	O
rapidly	O
.	O
</s>
<s>
This	O
suggests	O
that	O
,	O
even	O
after	O
knowing	O
completely	O
the	O
value	O
of	O
a	O
tag	B-General_Concept
,	O
the	O
entropy	O
of	O
the	O
set	O
of	O
documents	O
is	O
increasing	O
over	O
time	O
.	O
</s>
<s>
Conditional	O
entropy	O
asks	O
the	O
question	O
:	O
"	O
Given	O
that	O
a	O
set	O
of	O
tags	B-General_Concept
is	O
known	O
,	O
how	O
much	O
uncertainty	O
remains	O
regarding	O
the	O
document	O
set	O
referenced	O
by	O
those	O
tags	B-General_Concept
?	O
"	O
</s>
<s>
This	O
curve	O
is	O
strictly	O
increasing	O
,	O
which	O
suggests	O
that	O
the	O
specificity	O
of	O
any	O
given	O
tag	B-General_Concept
is	O
decreasing	O
.	O
</s>
<s>
As	O
a	O
navigation	O
aid	O
,	O
tags	B-General_Concept
are	O
becoming	O
harder	O
and	O
harder	O
to	O
use	O
,	O
and	O
a	O
single	O
tag	B-General_Concept
will	O
gradually	O
reference	O
too	O
many	O
documents	O
to	O
be	O
considered	O
useful	O
.	O
</s>
<s>
Chi	O
and	O
Mytkowicz	O
's	O
research	O
shows	O
that	O
as	O
a	O
measure	O
of	O
usefulness	O
of	O
tags	B-General_Concept
and	O
their	O
encoding	O
,	O
there	O
is	O
a	O
worsening	O
trend	O
in	O
the	O
ability	O
of	O
users	O
to	O
specify	O
and	O
find	O
tags	B-General_Concept
and	O
documents	O
when	O
they	O
are	O
engaged	O
in	O
simple	O
fact	O
retrieval	O
.	O
</s>
<s>
This	O
suggests	O
that	O
search	O
and	O
recommendation	B-Application
systems	I-Application
should	O
be	O
built	O
to	O
help	O
users	O
sift	O
through	O
resources	O
in	O
social	B-General_Concept
tagging	I-General_Concept
systems	O
,	O
especially	O
when	O
they	O
are	O
engaged	O
in	O
activities	O
beyond	O
fact	O
retrieval	O
,	O
as	O
characterized	O
by	O
information	O
theory	O
.	O
</s>
<s>
Although	O
the	O
number	O
of	O
documents	O
associated	O
with	O
any	O
given	O
tag	B-General_Concept
is	O
increasing	O
,	O
there	O
are	O
many	O
ways	O
contextual	O
information	O
can	O
help	O
users	O
to	O
look	O
for	O
relevant	O
information	O
.	O
</s>
<s>
This	O
is	O
one	O
of	O
the	O
major	O
weakness	O
of	O
the	O
simple	O
information	O
theory	O
in	O
explaining	O
usefulness	O
of	O
tags	B-General_Concept
–	O
it	O
ignores	O
how	O
humans	O
can	O
extract	O
meanings	O
from	O
a	O
set	O
of	O
tags	B-General_Concept
assigned	O
to	O
a	O
document	O
.	O
</s>
<s>
For	O
example	O
,	O
a	O
2007	O
paper	O
showed	O
that	O
while	O
the	O
number	O
of	O
tags	B-General_Concept
is	O
increasing	O
,	O
the	O
general	O
growth	O
pattern	O
is	O
scale-free	O
–	O
the	O
general	O
distribution	O
of	O
tag-tag	O
co-occurrences	O
follows	O
a	O
power	O
law	O
.	O
</s>
<s>
The	O
same	O
paper	O
also	O
found	O
that	O
the	O
characteristics	O
of	O
this	O
scale-free	O
distribution	O
depend	O
on	O
the	O
semantics	B-Application
of	O
the	O
tag	B-General_Concept
–	O
tags	B-General_Concept
that	O
are	O
semantically	O
general	O
(	O
e.g.	O
,	O
blogs	O
)	O
tend	O
to	O
co-occur	O
with	O
many	O
tags	B-General_Concept
,	O
while	O
semantically	O
narrow	O
tags	B-General_Concept
(	O
e.g.	O
,	O
Ajax	B-General_Concept
)	O
tend	O
to	O
co-occur	O
with	O
few	O
tags	B-General_Concept
across	O
a	O
wide	O
set	O
of	O
documents	O
in	O
a	O
social	B-General_Concept
tagging	I-General_Concept
system	O
.	O
</s>
<s>
This	O
suggests	O
that	O
the	O
assumption	O
of	O
the	O
information	O
theory	O
approach	O
is	O
too	O
simple	O
–	O
when	O
taking	O
into	O
account	O
the	O
semantics	B-Application
of	O
the	O
set	O
of	O
tags	B-General_Concept
assigned	O
to	O
documents	O
,	O
the	O
predictive	O
value	O
of	O
tags	B-General_Concept
on	O
contents	O
of	O
documents	O
is	O
relatively	O
stable	O
.	O
</s>
<s>
This	O
finding	O
is	O
important	O
for	O
development	O
of	O
recommender	B-Application
systems	I-Application
–	O
discovering	O
these	O
higher-level	O
semantic	O
patterns	O
is	O
important	O
in	O
helping	O
people	O
find	O
relevant	O
information	O
.	O
</s>
<s>
Despite	O
this	O
potential	O
vocabulary	O
problem	O
,	O
research	O
has	O
found	O
that	O
at	O
the	O
aggregate	O
level	O
,	O
tagging	O
behavior	O
seemed	O
relatively	O
stable	O
,	O
and	O
that	O
the	O
tag	B-General_Concept
choice	O
proportions	O
seemed	O
to	O
be	O
converging	O
rather	O
than	O
diverging	O
.	O
</s>
<s>
While	O
these	O
observations	O
provided	O
evidence	O
against	O
the	O
proposed	O
vocabulary	O
problem	O
,	O
they	O
also	O
initiated	O
research	O
investigating	O
how	O
and	O
why	O
tag	B-General_Concept
proportions	O
tended	O
to	O
converge	O
over	O
time	O
.	O
</s>
<s>
One	O
explanation	O
for	O
the	O
stability	O
was	O
that	O
there	O
was	O
an	O
inherent	O
propensity	O
for	O
users	O
to	O
"	O
imitate	O
"	O
word	O
use	O
of	O
others	O
as	O
they	O
create	O
tags	B-General_Concept
.	O
</s>
<s>
This	O
propensity	O
may	O
act	O
as	O
a	O
form	O
of	O
social	O
cohesion	O
that	O
fosters	O
the	O
coherence	O
of	O
tag-topic	O
relations	O
in	O
the	O
system	O
,	O
and	O
leads	O
to	O
stability	O
in	O
the	O
system	O
.	O
</s>
<s>
It	O
was	O
shown	O
that	O
the	O
stochastic	O
urn	O
model	O
created	O
in	O
1923	O
was	O
useful	O
in	O
explaining	O
how	O
simple	O
imitation	O
behavior	O
at	O
the	O
individual	O
level	O
could	O
explain	O
the	O
converging	O
usage	O
patterns	O
of	O
tags	B-General_Concept
.	O
</s>
<s>
Specifically	O
,	O
the	O
convergence	O
of	O
tag	B-General_Concept
choices	O
was	O
simulated	O
by	O
a	O
process	O
in	O
which	O
a	O
colored	O
ball	O
was	O
randomly	O
selected	O
from	O
an	O
urn	O
,	O
then	O
replaced	O
in	O
the	O
urn	O
along	O
with	O
an	O
additional	O
ball	O
of	O
the	O
same	O
color	O
,	O
simulating	O
the	O
probabilistic	O
nature	O
of	O
tag	B-General_Concept
reuse	O
.	O
</s>
<s>
This	O
simple	O
model	O
,	O
however	O
,	O
does	O
not	O
explain	O
why	O
certain	O
tags	B-General_Concept
would	O
be	O
"	O
imitated	O
"	O
more	O
often	O
than	O
others	O
,	O
and	O
therefore	O
cannot	O
provide	O
a	O
realistic	O
mechanism	O
for	O
tag	B-General_Concept
choices	O
and	O
how	O
social	O
tags	B-General_Concept
could	O
be	O
used	O
as	O
navigational	O
cues	O
during	O
exploratory	O
search	O
.	O
</s>
<s>
Research	O
based	O
on	O
data	O
from	O
the	O
social	O
bookmarking	O
website	O
Del.icio.us	B-Application
has	O
shown	O
that	O
collaborative	B-General_Concept
tagging	I-General_Concept
systems	O
exhibit	O
a	O
form	O
of	O
complex	O
systems	O
(	O
or	O
self-organizing	O
)	O
dynamics	O
.	O
</s>
<s>
Furthermore	O
,	O
although	O
there	O
is	O
no	O
central	O
,	O
controlled	O
vocabulary	O
to	O
constrain	O
the	O
actions	O
of	O
individual	O
users	O
,	O
the	O
distributions	O
of	O
tags	B-General_Concept
that	O
describe	O
different	O
resources	O
has	O
been	O
shown	O
to	O
converge	O
over	O
time	O
to	O
a	O
stable	O
,	O
power-law	O
distribution	O
.	O
</s>
<s>
Once	O
such	O
stable	O
distributions	O
form	O
,	O
the	O
correlations	O
between	O
different	O
tags	B-General_Concept
can	O
be	O
used	O
to	O
construct	O
simple	O
folksonomy	B-General_Concept
graphs	O
,	O
which	O
can	O
be	O
partitioned	O
to	O
obtain	O
a	O
form	O
of	O
community	O
or	O
shared	O
vocabularies	O
.	O
</s>
<s>
The	O
memory-based	O
Yule-Simon	O
(	O
MBYS	O
)	O
model	O
attempts	O
to	O
explain	O
tag	B-General_Concept
choices	O
by	O
a	O
stochastic	O
process	O
.	O
</s>
<s>
It	O
was	O
found	O
that	O
the	O
temporal	O
order	O
of	O
tag	B-General_Concept
assignment	O
influences	O
users	O
 '	O
tag	B-General_Concept
choices	O
.	O
</s>
<s>
Similar	O
to	O
the	O
stochastic	O
urn	O
model	O
,	O
the	O
MBYS	O
model	O
assumes	O
that	O
at	O
each	O
step	O
,	O
a	O
tag	B-General_Concept
would	O
be	O
randomly	O
sampled	O
:	O
with	O
probability	O
that	O
the	O
sampled	O
tag	B-General_Concept
was	O
new	O
,	O
and	O
with	O
probability	O
1	O
-	O
that	O
the	O
sampled	O
tag	B-General_Concept
was	O
copied	O
from	O
existing	O
tags	B-General_Concept
.	O
</s>
<s>
When	O
copying	O
,	O
the	O
probability	O
of	O
selecting	O
a	O
tag	B-General_Concept
was	O
assumed	O
to	O
decay	O
with	O
time	O
,	O
and	O
this	O
decay	O
function	O
was	O
found	O
to	O
follow	O
a	O
power-law	O
distribution	O
.	O
</s>
<s>
Thus	O
,	O
tags	B-General_Concept
that	O
were	O
more	O
recently	O
used	O
had	O
a	O
higher	O
probability	O
of	O
being	O
reused	O
.	O
</s>
<s>
One	O
major	O
finding	O
was	O
that	O
semantically	O
general	O
tags	B-General_Concept
(	O
e.g.	O
,	O
"	O
blog	O
"	O
)	O
generally	O
co-occurred	O
more	O
frequently	O
with	O
other	O
tags	B-General_Concept
than	O
semantically	O
narrower	O
tags	B-General_Concept
(	O
e.g.	O
,	O
"	O
Ajax	B-General_Concept
"	O
)	O
,	O
and	O
this	O
difference	O
could	O
be	O
captured	O
by	O
the	O
decay	O
function	O
of	O
tag	B-General_Concept
reuse	O
in	O
their	O
model	O
.	O
</s>
<s>
Specifically	O
,	O
it	O
was	O
found	O
that	O
a	O
slower	O
decay	O
parameter	O
(	O
when	O
the	O
tag	B-General_Concept
is	O
reused	O
more	O
often	O
)	O
could	O
explain	O
the	O
phenomenon	O
that	O
semantically	O
general	O
tags	B-General_Concept
tended	O
to	O
co-occur	O
with	O
a	O
larger	O
set	O
of	O
tags	B-General_Concept
.	O
</s>
<s>
In	O
other	O
words	O
,	O
it	O
was	O
argued	O
that	O
the	O
"	O
semantic	O
breadth	O
"	O
of	O
a	O
tag	B-General_Concept
could	O
be	O
modeled	O
by	O
a	O
memory	O
decay	O
function	O
,	O
which	O
could	O
lead	O
to	O
different	O
emergent	O
behavioral	O
patterns	O
in	O
a	O
tagging	O
system	O
.	O
</s>
<s>
Descriptive	B-General_Concept
models	O
were	O
based	O
on	O
analyses	O
of	O
word-word	O
relations	O
as	O
revealed	O
by	O
the	O
various	O
statistical	O
structures	O
in	O
the	O
organization	O
of	O
tags	B-General_Concept
(	O
e.g.	O
,	O
how	O
likely	O
one	O
tag	B-General_Concept
would	O
co-occur	O
with	O
other	O
tags	B-General_Concept
or	O
how	O
likely	O
each	O
tag	B-General_Concept
was	O
reused	O
over	O
time	O
)	O
.	O
</s>
<s>
Thus	O
,	O
these	O
models	O
are	O
descriptive	B-General_Concept
models	O
at	O
the	O
aggregate	O
level	O
,	O
and	O
have	O
little	O
to	O
offer	O
about	O
predictions	O
at	O
the	O
level	O
of	O
an	O
individual	O
's	O
interface	O
interactions	O
and	O
cognitive	O
processes	O
.	O
</s>
<s>
Rather	O
than	O
imitating	O
other	O
users	O
at	O
the	O
word	O
level	O
,	O
one	O
possible	O
explanation	O
for	O
this	O
kind	O
of	O
social	O
cohesion	O
could	O
be	O
grounded	O
on	O
the	O
natural	O
tendency	O
for	O
people	O
to	O
process	O
tags	B-General_Concept
at	O
the	O
semantic	O
level	O
,	O
and	O
it	O
was	O
at	O
this	O
level	O
of	O
processing	O
that	O
most	O
of	O
the	O
imitation	O
occurred	O
.	O
</s>
<s>
Assuming	O
that	O
people	O
in	O
the	O
same	O
culture	O
tend	O
to	O
have	O
shared	O
structures	O
–	O
such	O
as	O
using	O
similar	O
vocabularies	O
and	O
their	O
corresponding	O
meanings	O
to	O
conform	O
and	O
communicate	O
,	O
users	O
of	O
the	O
same	O
social	B-General_Concept
tagging	I-General_Concept
system	O
may	O
also	O
share	O
similar	O
semantic	O
representations	O
of	O
words	O
and	O
concepts	O
,	O
even	O
when	O
the	O
use	O
of	O
tags	B-General_Concept
may	O
vary	O
across	O
individuals	O
at	O
the	O
word	O
level	O
.	O
</s>
<s>
As	O
such	O
,	O
part	O
of	O
the	O
reason	O
for	O
the	O
stability	O
of	O
social	B-General_Concept
tagging	I-General_Concept
systems	O
can	O
be	O
attributed	O
to	O
the	O
shared	O
semantic	O
representations	O
among	O
the	O
users	O
,	O
such	O
that	O
users	O
may	O
have	O
relatively	O
stable	O
and	O
coherent	O
interpretation	O
of	O
information	O
contents	O
and	O
tags	B-General_Concept
as	O
they	O
interact	O
with	O
the	O
system	O
.	O
</s>
<s>
Based	O
on	O
this	O
assumption	O
,	O
the	O
semantic	O
imitation	O
model	O
predicts	O
how	O
different	O
semantic	O
representations	O
may	O
lead	O
to	O
differences	O
in	O
individual	O
tag	B-General_Concept
choices	O
and	O
eventually	O
different	O
emergent	O
properties	O
at	O
the	O
aggregate	O
behavioral	O
level	O
.	O
</s>
<s>
The	O
model	O
also	O
predicts	O
that	O
the	O
folksonomies	B-General_Concept
in	O
the	O
system	O
reflect	O
the	O
shared	O
semantic	O
representations	O
of	O
the	O
users	O
.	O
</s>
<s>
Semantic	O
imitation	O
has	O
important	O
implications	O
to	O
the	O
general	O
vocabulary	O
problem	O
in	O
information	O
retrieval	O
and	O
human	O
–	O
computer	O
interaction	O
–	O
the	O
creation	O
of	O
a	O
large	O
number	O
of	O
diverse	O
tags	B-General_Concept
to	O
describe	O
the	O
same	O
set	O
of	O
information	O
resources	O
.	O
</s>
<s>
Thus	O
,	O
although	O
there	O
may	O
not	O
be	O
strong	O
coherence	O
in	O
the	O
choice	O
of	O
words	O
in	O
describing	O
a	O
resource	O
,	O
at	O
the	O
semantic	O
level	O
,	O
there	O
seems	O
to	O
be	O
a	O
stronger	O
coherence	O
force	O
that	O
guides	O
the	O
convergence	O
of	O
descriptive	B-General_Concept
indices	O
.	O
</s>
<s>
Instead	O
,	O
the	O
process	O
of	O
semantic	O
imitation	O
in	O
social	B-General_Concept
tagging	I-General_Concept
implies	O
that	O
the	O
information-theoretic	O
approach	O
is	O
at	O
most	O
incomplete	O
,	O
as	O
it	O
does	O
not	O
take	O
into	O
account	O
the	O
basic	O
unit	O
of	O
human	O
information	O
processing	O
.	O
</s>
<s>
In	O
the	O
social	B-General_Concept
tagging	I-General_Concept
case	O
,	O
as	O
long	O
as	O
users	O
share	O
a	O
similar	O
understanding	O
of	O
the	O
contents	O
of	O
the	O
information	O
resources	O
,	O
the	O
fact	O
that	O
the	O
information	O
value	O
of	O
tag-document	O
decreases	O
(	O
that	O
humans	O
have	O
more	O
words	O
in	O
their	O
languages	O
)	O
does	O
not	O
imply	O
that	O
it	O
will	O
always	O
be	O
harder	O
to	O
find	O
relevant	O
information	O
(	O
similarly	O
,	O
the	O
fact	O
that	O
there	O
are	O
an	O
increasing	O
number	O
words	O
in	O
human	O
languages	O
does	O
not	O
mean	O
that	O
communication	O
becomes	O
less	O
effective	O
)	O
.	O
</s>
<s>
However	O
,	O
it	O
does	O
point	O
to	O
the	O
notion	O
that	O
one	O
needs	O
to	O
effectively	O
present	O
these	O
semantic	O
structures	O
in	O
the	O
information	O
system	O
so	O
that	O
people	O
can	O
effectively	O
interpret	O
the	O
semantics	B-Application
of	O
the	O
tagged	O
documents	O
.	O
</s>
