<s>
Noisy	B-General_Concept
text	I-General_Concept
analytics	I-General_Concept
is	O
a	O
process	O
of	O
information	B-General_Concept
extraction	I-General_Concept
whose	O
goal	O
is	O
to	O
automatically	O
extract	O
structured	O
or	O
semistructured	O
information	O
from	O
noisy	O
unstructured	O
text	O
data	O
.	O
</s>
<s>
While	O
Text	B-Algorithm
analytics	I-Algorithm
is	O
a	O
growing	O
and	O
mature	O
field	O
that	O
has	O
great	O
value	O
because	O
of	O
the	O
huge	O
amounts	O
of	O
data	O
being	O
produced	O
,	O
processing	O
of	O
noisy	O
text	O
is	O
gaining	O
in	O
importance	O
because	O
a	O
lot	O
of	O
common	O
applications	O
produce	O
noisy	O
text	O
data	O
.	O
</s>
<s>
Noisy	O
unstructured	O
text	O
data	O
is	O
found	O
in	O
informal	O
settings	O
such	O
as	O
online	B-Protocol
chat	I-Protocol
,	O
text	O
messages	O
,	O
e-mails	O
,	O
message	B-Application
boards	I-Application
,	O
newsgroups	B-Application
,	O
blogs	O
,	O
wikis	B-Protocol
and	O
web	O
pages	O
.	O
</s>
<s>
Also	O
,	O
text	O
produced	O
by	O
processing	O
spontaneous	O
speech	O
using	O
automatic	B-Application
speech	I-Application
recognition	I-Application
and	O
printed	O
or	O
handwritten	O
text	O
using	O
optical	B-Application
character	I-Application
recognition	I-Application
contains	O
processing	O
noise	O
.	O
</s>
<s>
Such	O
text	O
can	O
be	O
seen	O
in	O
large	O
amounts	O
in	O
contact	O
centers	O
,	O
chat	B-Protocol
rooms	O
,	O
optical	B-Application
character	I-Application
recognition	I-Application
(	O
OCR	O
)	O
of	O
text	O
documents	O
,	O
short	O
message	O
service	O
(	O
SMS	O
)	O
text	O
,	O
etc	O
.	O
</s>
<s>
and	O
parsing	B-Language
.	O
</s>
<s>
World	O
Wide	O
Web	O
:	O
Poorly	O
written	O
text	O
is	O
found	O
in	O
web	O
pages	O
,	O
online	B-Protocol
chat	I-Protocol
,	O
blogs	O
,	O
wikis	B-Protocol
,	O
discussion	B-Application
forums	I-Application
,	O
newsgroups	B-Application
.	O
</s>
<s>
Also	O
,	O
because	O
of	O
the	O
large	O
amount	O
of	O
data	O
,	O
it	O
is	O
necessary	O
to	O
find	O
efficient	O
methods	O
of	O
information	B-General_Concept
extraction	I-General_Concept
,	O
classification	B-General_Concept
,	O
automatic	B-Application
summarization	I-Application
and	O
analysis	O
of	O
these	O
data	O
.	O
</s>
<s>
They	O
operate	O
in	O
various	O
modes	O
such	O
as	O
voice	O
,	O
online	B-Protocol
chat	I-Protocol
and	O
E-mail	O
.	O
</s>
<s>
The	O
contact	O
center	O
industry	O
produces	O
gigabytes	O
of	O
data	O
in	O
the	O
form	O
of	O
E-mails	O
,	O
chat	B-Protocol
logs	O
,	O
voice	O
conversation	O
transcriptions	O
,	O
customer	O
feedback	O
,	O
etc	O
.	O
</s>
<s>
Transcription	O
of	O
these	O
using	O
state	O
of	O
the	O
art	O
automatic	B-Application
speech	I-Application
recognition	I-Application
results	O
in	O
text	O
with	O
30-40	O
%	O
word	B-General_Concept
error	I-General_Concept
rate	I-General_Concept
.	O
</s>
<s>
Further	O
,	O
even	O
written	O
modes	O
of	O
communication	O
like	O
online	B-Protocol
chat	I-Protocol
between	O
customers	O
and	O
agents	O
and	O
even	O
the	O
interactions	O
over	O
email	O
tend	O
to	O
be	O
noisy	O
.	O
</s>
<s>
Printed	O
Documents	O
:	O
Many	O
libraries	O
,	O
government	O
organizations	O
and	O
national	O
defence	O
organizations	O
have	O
vast	O
repositories	O
of	O
hard	B-General_Concept
copy	I-General_Concept
documents	O
.	O
</s>
<s>
To	O
retrieve	O
and	O
process	O
the	O
content	O
from	O
such	O
documents	O
,	O
they	O
need	O
to	O
be	O
processed	O
using	O
Optical	B-Application
Character	I-Application
Recognition	I-Application
.	O
</s>
<s>
It	O
can	O
range	O
from	O
2-3	O
%	O
word	B-General_Concept
error	I-General_Concept
rates	I-General_Concept
to	O
as	O
high	O
as	O
50-60	O
%	O
word	B-General_Concept
error	I-General_Concept
rates	I-General_Concept
.	O
</s>
<s>
Short	O
Messaging	O
Service	O
(	O
SMS	O
)	O
:	O
Language	O
usage	O
over	O
computer	O
mediated	O
discourses	O
,	O
like	O
chats	B-Protocol
,	O
emails	O
and	O
SMS	O
texts	O
,	O
significantly	O
differs	O
from	O
the	O
standard	O
form	O
of	O
the	O
language	O
.	O
</s>
