<s>
Big	B-Application
data	I-Application
primarily	O
refers	O
to	O
data	B-General_Concept
sets	I-General_Concept
that	O
are	O
too	O
large	O
or	O
complex	O
to	O
be	O
dealt	O
with	O
by	O
traditional	O
data-processing	B-General_Concept
application	B-Application
software	I-Application
.	O
</s>
<s>
Data	O
with	O
many	O
entries	O
(	O
rows	O
)	O
offer	O
greater	O
statistical	B-General_Concept
power	I-General_Concept
,	O
while	O
data	O
with	O
higher	O
complexity	O
(	O
more	O
attributes	O
or	O
columns	O
)	O
may	O
lead	O
to	O
a	O
higher	O
false	B-General_Concept
discovery	I-General_Concept
rate	I-General_Concept
.	O
</s>
<s>
Though	O
used	O
sometimes	O
loosely	O
partly	O
because	O
of	O
a	O
lack	O
of	O
formal	O
definition	O
,	O
the	O
interpretation	O
that	O
seems	O
to	O
best	O
describe	O
big	B-Application
data	I-Application
is	O
the	O
one	O
associated	O
with	O
large	O
body	O
of	O
information	O
that	O
we	O
could	O
not	O
comprehend	O
when	O
used	O
only	O
in	O
smaller	O
amounts	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
analysis	I-Application
challenges	O
include	O
capturing	B-General_Concept
data	I-General_Concept
,	O
data	B-General_Concept
storage	I-General_Concept
,	O
data	B-General_Concept
analysis	I-General_Concept
,	O
search	O
,	O
sharing	B-Application
,	O
transfer	O
,	O
visualization	B-Application
,	O
querying	B-Language
,	O
updating	O
,	O
information	O
privacy	O
,	O
and	O
data	O
source	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
was	O
originally	O
associated	O
with	O
three	O
key	O
concepts	O
:	O
volume	O
,	O
variety	O
,	O
and	O
velocity	O
.	O
</s>
<s>
The	O
analysis	O
of	O
big	B-Application
data	I-Application
presents	O
challenges	O
in	O
sampling	O
,	O
and	O
thus	O
previously	O
allowing	O
for	O
only	O
observations	O
and	O
sampling	O
.	O
</s>
<s>
Without	O
sufficient	O
investment	O
in	O
expertise	O
for	O
big	B-Application
data	I-Application
veracity	O
,	O
then	O
the	O
volume	O
and	O
variety	O
of	O
data	O
can	O
produce	O
costs	O
and	O
risks	O
that	O
exceed	O
an	O
organization	O
's	O
capacity	O
to	O
create	O
and	O
capture	B-Algorithm
value	O
from	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Current	O
usage	O
of	O
the	O
term	O
big	B-Application
data	I-Application
tends	O
to	O
refer	O
to	O
the	O
use	O
of	O
predictive	B-General_Concept
analytics	I-General_Concept
,	O
user	B-Application
behavior	I-Application
analytics	I-Application
,	O
or	O
certain	O
other	O
advanced	O
data	B-General_Concept
analytics	I-General_Concept
methods	O
that	O
extract	O
value	O
from	O
big	B-Application
data	I-Application
,	O
and	O
seldom	O
to	O
a	O
particular	O
size	O
of	O
data	B-General_Concept
set	I-General_Concept
.	O
</s>
<s>
Analysis	O
of	O
data	B-General_Concept
sets	I-General_Concept
can	O
find	O
new	O
correlations	O
to	O
"	O
spot	O
business	O
trends	O
,	O
prevent	O
diseases	O
,	O
combat	O
crime	O
and	O
so	O
on	O
"	O
.	O
</s>
<s>
Scientists	O
,	O
business	O
executives	O
,	O
medical	O
practitioners	O
,	O
advertising	O
and	O
governments	O
alike	O
regularly	O
meet	O
difficulties	O
with	O
large	O
data-sets	B-General_Concept
in	O
areas	O
including	O
Internet	B-Application
searches	I-Application
,	O
fintech	O
,	O
healthcare	O
analytics	O
,	O
geographic	O
information	O
systems	O
,	O
urban	O
informatics	O
,	O
and	O
business	B-Application
informatics	I-Application
.	O
</s>
<s>
The	O
size	O
and	O
number	O
of	O
available	O
data	B-General_Concept
sets	I-General_Concept
have	O
grown	O
rapidly	O
as	O
data	O
is	O
collected	O
by	O
devices	O
such	O
as	O
mobile	B-Application
devices	I-Application
,	O
cheap	O
and	O
numerous	O
information-sensing	O
Internet	B-Operating_System
of	I-Operating_System
things	I-Operating_System
devices	O
,	O
aerial	O
(	O
remote	B-Application
sensing	I-Application
)	O
,	O
software	O
logs	O
,	O
cameras	B-Device
,	O
microphones	O
,	O
radio-frequency	B-Application
identification	I-Application
(	O
RFID	B-Application
)	O
readers	O
and	O
wireless	B-Architecture
sensor	I-Architecture
networks	I-Architecture
.	O
</s>
<s>
According	O
to	O
IDC	O
,	O
global	O
spending	O
on	O
big	B-Application
data	I-Application
and	O
business	O
analytics	O
(	O
BDA	O
)	O
solutions	O
is	O
estimated	O
to	O
reach	O
$215.7	O
billion	O
in	O
2021	O
.	O
</s>
<s>
While	O
Statista	O
report	O
,	O
the	O
global	O
big	B-Application
data	I-Application
market	O
is	O
forecasted	O
to	O
grow	O
to	O
$103	O
billion	O
by	O
2027	O
.	O
</s>
<s>
In	O
2011	O
McKinsey	O
&	O
Company	O
reported	O
,	O
if	O
US	O
healthcare	O
were	O
to	O
use	O
big	B-Application
data	I-Application
creatively	O
and	O
effectively	O
to	O
drive	O
efficiency	O
and	O
quality	O
,	O
the	O
sector	O
could	O
create	O
more	O
than	O
$300	O
billion	O
in	O
value	O
every	O
year	O
.	O
</s>
<s>
In	O
the	O
developed	O
economies	O
of	O
Europe	O
,	O
government	O
administrators	O
could	O
save	O
more	O
than	O
€100	O
billion	O
( $149	O
billion	O
)	O
in	O
operational	O
efficiency	O
improvements	O
alone	O
by	O
using	O
big	B-Application
data	I-Application
.	O
</s>
<s>
And	O
users	O
of	O
services	O
enabled	O
by	O
personal-location	O
data	O
could	O
capture	B-Algorithm
$600	O
billion	O
in	O
consumer	O
surplus	O
.	O
</s>
<s>
One	O
question	O
for	O
large	O
enterprises	O
is	O
determining	O
who	O
should	O
own	O
big-data	B-Application
initiatives	O
that	O
affect	O
the	O
entire	O
organization	O
.	O
</s>
<s>
Relational	B-Application
database	I-Application
management	I-Application
systems	I-Application
and	O
desktop	O
statistical	O
software	O
packages	O
used	O
to	O
visualize	O
data	O
often	O
have	O
difficulty	O
processing	O
and	O
analyzing	O
big	B-Application
data	I-Application
.	O
</s>
<s>
The	O
processing	O
and	O
analysis	O
of	O
big	B-Application
data	I-Application
may	O
require	O
"	O
massively	B-Operating_System
parallel	I-Operating_System
software	O
running	O
on	O
tens	O
,	O
hundreds	O
,	O
or	O
even	O
thousands	O
of	O
servers	O
"	O
.	O
</s>
<s>
What	O
qualifies	O
as	O
"	O
big	B-Application
data	I-Application
"	O
varies	O
depending	O
on	O
the	O
capabilities	O
of	O
those	O
analyzing	O
it	O
and	O
their	O
tools	O
.	O
</s>
<s>
Furthermore	O
,	O
expanding	O
capabilities	O
make	O
big	B-Application
data	I-Application
a	O
moving	O
target	O
.	O
</s>
<s>
The	O
term	O
big	B-Application
data	I-Application
has	O
been	O
in	O
use	O
since	O
the	O
1990s	O
,	O
with	O
some	O
giving	O
credit	O
to	O
John	O
Mashey	O
for	O
popularizing	O
the	O
term	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
usually	O
includes	O
data	B-General_Concept
sets	I-General_Concept
with	O
sizes	O
beyond	O
the	O
ability	O
of	O
commonly	O
used	O
software	O
tools	O
to	O
capture	B-Algorithm
,	O
curate	O
,	O
manage	O
,	O
and	O
process	O
data	O
within	O
a	O
tolerable	O
elapsed	O
time	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
philosophy	O
encompasses	O
unstructured	O
,	O
semi-structured	O
and	O
structured	O
data	O
;	O
however	O
,	O
the	O
main	O
focus	O
is	O
on	O
unstructured	O
data	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
"	O
size	O
"	O
is	O
a	O
constantly	O
moving	O
target	O
;	O
ranging	O
from	O
a	O
few	O
dozen	O
terabytes	O
to	O
many	O
zettabytes	O
of	O
data	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
requires	O
a	O
set	O
of	O
techniques	O
and	O
technologies	O
with	O
new	O
forms	O
of	O
integration	B-General_Concept
to	O
reveal	O
insights	O
from	O
data-sets	B-General_Concept
that	O
are	O
diverse	O
,	O
complex	O
,	O
and	O
of	O
a	O
massive	O
scale	O
.	O
</s>
<s>
The	O
Vs	O
of	O
big	B-Application
data	I-Application
were	O
often	O
referred	O
to	O
as	O
the	O
"	O
three	O
Vs	O
"	O
,	O
"	O
four	O
Vs	O
"	O
,	O
and	O
"	O
five	O
Vs	O
"	O
.	O
</s>
<s>
They	O
represented	O
the	O
qualities	O
of	O
big	B-Application
data	I-Application
in	O
volume	O
,	O
variety	O
,	O
velocity	O
,	O
veracity	O
,	O
and	O
value	O
.	O
</s>
<s>
Variability	O
is	O
often	O
included	O
as	O
an	O
additional	O
quality	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
A	O
2018	O
definition	O
states	O
"	O
Big	B-Application
data	I-Application
is	O
where	O
parallel	O
computing	O
tools	O
are	O
needed	O
to	O
handle	O
data	O
"	O
,	O
and	O
notes	O
,	O
"	O
This	O
represents	O
a	O
distinct	O
and	O
clearly	O
defined	O
change	O
in	O
the	O
computer	O
science	O
used	O
,	O
via	O
parallel	O
programming	O
theories	O
,	O
and	O
losses	O
of	O
some	O
of	O
the	O
guarantees	O
and	O
capabilities	O
made	O
by	O
Codd	O
's	O
relational	O
model.	O
"	O
</s>
<s>
In	O
a	O
comparative	O
study	O
of	O
big	O
datasets	B-General_Concept
,	O
Kitchin	O
and	O
McArdle	O
found	O
that	O
none	O
of	O
the	O
commonly	O
considered	O
characteristics	O
of	O
big	B-Application
data	I-Application
appear	O
consistently	O
across	O
all	O
of	O
the	O
analyzed	O
cases	O
.	O
</s>
<s>
Instead	O
of	O
focusing	O
on	O
intrinsic	O
characteristics	O
of	O
big	B-Application
data	I-Application
,	O
this	O
alternative	O
perspective	O
pushes	O
forward	O
a	O
relational	O
understanding	O
of	O
the	O
object	O
claiming	O
that	O
what	O
matters	O
is	O
the	O
way	O
in	O
which	O
data	O
is	O
collected	O
,	O
stored	O
,	O
made	O
available	O
and	O
analyzed	O
.	O
</s>
<s>
The	O
growing	O
maturity	O
of	O
the	O
concept	O
more	O
starkly	O
delineates	O
the	O
difference	O
between	O
"	O
big	B-Application
data	I-Application
"	O
and	O
"	O
business	B-General_Concept
intelligence	I-General_Concept
"	O
:	O
</s>
<s>
Business	B-General_Concept
intelligence	I-General_Concept
uses	O
applied	O
mathematics	O
tools	O
and	O
descriptive	B-General_Concept
statistics	I-General_Concept
with	O
data	O
with	O
high	O
information	O
density	O
to	O
measure	O
things	O
,	O
detect	O
trends	O
,	O
etc	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
uses	O
mathematical	O
analysis	O
,	O
optimization	O
,	O
inductive	O
statistics	O
,	O
and	O
concepts	O
from	O
nonlinear	O
system	O
identification	O
to	O
infer	O
laws	O
(	O
regressions	O
,	O
nonlinear	O
relationships	O
,	O
and	O
causal	O
effects	O
)	O
from	O
large	O
sets	O
of	O
data	O
with	O
low	O
information	O
density	O
to	O
reveal	O
relationships	O
and	O
dependencies	O
,	O
or	O
to	O
perform	O
predictions	O
of	O
outcomes	O
and	O
behaviors	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
can	O
be	O
described	O
by	O
the	O
following	O
characteristics	O
:	O
</s>
<s>
The	O
size	O
of	O
the	O
data	O
determines	O
the	O
value	O
and	O
potential	O
insight	O
,	O
and	O
whether	O
it	O
can	O
be	O
considered	O
big	B-Application
data	I-Application
or	O
not	O
.	O
</s>
<s>
The	O
size	O
of	O
big	B-Application
data	I-Application
is	O
usually	O
larger	O
than	O
terabytes	O
and	O
petabytes	O
.	O
</s>
<s>
The	O
big	B-Application
data	I-Application
technologies	O
evolved	O
with	O
the	O
prime	O
intention	O
to	O
capture	B-Algorithm
,	O
store	O
,	O
and	O
process	O
the	O
semi-structured	O
and	O
unstructured	O
(	O
variety	O
)	O
data	O
generated	O
with	O
high	O
speed	O
(	O
velocity	O
)	O
,	O
and	O
huge	O
in	O
size	O
(	O
volume	O
)	O
.	O
</s>
<s>
Eventually	O
,	O
the	O
processing	O
of	O
structured	O
data	O
was	O
still	O
kept	O
as	O
optional	O
,	O
either	O
using	O
big	B-Application
data	I-Application
or	O
traditional	O
RDBMSs	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
draws	O
from	O
text	O
,	O
images	O
,	O
audio	O
,	O
video	O
;	O
plus	O
it	O
completes	O
missing	O
pieces	O
through	O
data	B-General_Concept
fusion	I-General_Concept
.	O
</s>
<s>
Big	B-Application
data	I-Application
is	O
often	O
available	O
in	O
real-time	O
.	O
</s>
<s>
Compared	O
to	O
small	B-General_Concept
data	I-General_Concept
,	O
big	B-Application
data	I-Application
is	O
produced	O
more	O
continually	O
.	O
</s>
<s>
Two	O
kinds	O
of	O
velocity	O
related	O
to	O
big	B-Application
data	I-Application
are	O
the	O
frequency	O
of	O
generation	O
and	O
the	O
frequency	O
of	O
handling	O
,	O
recording	O
,	O
and	O
publishing	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
must	O
not	O
only	O
be	O
large	O
in	O
size	O
,	O
but	O
also	O
must	O
be	O
reliable	O
in	O
order	O
to	O
achieve	O
value	O
in	O
the	O
analysis	O
of	O
it	O
.	O
</s>
<s>
Value	O
The	O
worth	O
in	O
information	O
that	O
can	O
be	O
achieved	O
by	O
the	O
processing	O
and	O
analysis	O
of	O
large	O
datasets	B-General_Concept
.	O
</s>
<s>
Value	O
also	O
can	O
be	O
measured	O
by	O
an	O
assessment	O
of	O
the	O
other	O
qualities	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Value	O
may	O
also	O
represent	O
the	O
profitability	O
of	O
information	O
that	O
is	O
retrieved	O
from	O
the	O
analysis	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Variability	O
The	O
characteristic	O
of	O
the	O
changing	O
formats	O
,	O
structure	O
,	O
or	O
sources	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Big	B-Application
data	I-Application
can	O
include	O
structured	O
,	O
unstructured	O
,	O
or	O
combinations	O
of	O
structured	O
and	O
unstructured	O
data	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
analysis	I-Application
may	O
integrate	O
raw	O
data	O
from	O
multiple	O
sources	O
.	O
</s>
<s>
Other	O
possible	O
characteristics	O
of	O
big	B-Application
data	I-Application
are	O
:	O
</s>
<s>
Big	B-Application
data	I-Application
may	O
or	O
may	O
not	O
include	O
all	O
the	O
available	O
data	O
from	O
sources	O
.	O
</s>
<s>
Relational	O
If	O
the	O
data	O
collected	O
contains	O
common	O
fields	O
that	O
would	O
enable	O
a	O
conjoining	O
,	O
or	O
meta-analysis	O
,	O
of	O
different	O
data	B-General_Concept
sets	I-General_Concept
.	O
</s>
<s>
Scalability	O
If	O
the	O
size	O
of	O
the	O
big	B-Application
data	I-Application
storage	B-General_Concept
system	I-General_Concept
can	O
expand	O
rapidly	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
repositories	O
have	O
existed	O
in	O
many	O
forms	O
,	O
often	O
built	O
by	O
corporations	O
with	O
a	O
special	O
need	O
.	O
</s>
<s>
Commercial	O
vendors	O
historically	O
offered	O
parallel	O
database	O
management	O
systems	O
for	O
big	B-Application
data	I-Application
beginning	O
in	O
the	O
1990s	O
.	O
</s>
<s>
Teradata	O
Corporation	O
in	O
1984	O
marketed	O
the	O
parallel	O
processing	O
DBC	B-Device
1012	I-Device
system	O
.	O
</s>
<s>
Hard	O
disk	O
drives	O
were	O
2.5	O
GB	O
in	O
1991	O
so	O
the	O
definition	O
of	O
big	B-Application
data	I-Application
continuously	O
evolves	O
.	O
</s>
<s>
Teradata	O
installed	O
the	O
first	O
petabyte	O
class	O
RDBMS	O
based	O
system	O
in	O
2007	O
.	O
,	O
there	O
are	O
a	O
few	O
dozen	O
petabyte	O
class	O
Teradata	O
relational	B-Application
databases	I-Application
installed	O
,	O
the	O
largest	O
of	O
which	O
exceeds	O
50	O
PB	O
.	O
</s>
<s>
Since	O
then	O
,	O
Teradata	O
has	O
added	O
unstructured	O
data	O
types	O
including	O
XML	B-Protocol
,	O
JSON	B-General_Concept
,	O
and	O
Avro	O
.	O
</s>
<s>
In	O
2000	O
,	O
Seisint	O
Inc	O
.	O
(	O
now	O
LexisNexis	O
Risk	O
Solutions	O
)	O
developed	O
a	O
C++	B-Language
-based	O
distributed	O
platform	O
for	O
data	B-General_Concept
processing	I-General_Concept
and	O
querying	B-Language
known	O
as	O
the	O
HPCC	O
Systems	O
platform	O
.	O
</s>
<s>
Users	O
can	O
write	O
data	B-General_Concept
processing	I-General_Concept
pipelines	O
and	O
queries	O
in	O
a	O
declarative	O
dataflow	O
programming	O
language	O
called	O
ECL	O
.	O
</s>
<s>
Data	B-General_Concept
analysts	I-General_Concept
working	O
in	O
ECL	O
are	O
not	O
required	O
to	O
define	O
data	O
schemas	O
upfront	O
and	O
can	O
rather	O
focus	O
on	O
the	O
particular	O
problem	O
at	O
hand	O
,	O
reshaping	O
data	O
in	O
the	O
best	O
possible	O
manner	O
as	O
they	O
develop	O
the	O
solution	O
.	O
</s>
<s>
CERN	O
and	O
other	O
physics	O
experiments	O
have	O
collected	O
big	B-Application
data	I-Application
sets	O
for	O
many	O
decades	O
,	O
usually	O
analyzed	O
via	O
high-throughput	B-Operating_System
computing	I-Operating_System
rather	O
than	O
the	O
map-reduce	B-Operating_System
architectures	O
usually	O
meant	O
by	O
the	O
current	O
"	O
big	B-Application
data	I-Application
"	O
movement	O
.	O
</s>
<s>
In	O
2004	O
,	O
Google	B-Application
published	O
a	O
paper	O
on	O
a	O
process	O
called	O
MapReduce	B-Operating_System
that	O
uses	O
a	O
similar	O
architecture	O
.	O
</s>
<s>
The	O
MapReduce	B-Operating_System
concept	O
provides	O
a	O
parallel	O
processing	O
model	O
,	O
and	O
an	O
associated	O
implementation	O
was	O
released	O
to	O
process	O
huge	O
amounts	O
of	O
data	O
.	O
</s>
<s>
With	O
MapReduce	B-Operating_System
,	O
queries	O
are	O
split	O
and	O
distributed	O
across	O
parallel	O
nodes	O
and	O
processed	O
in	O
parallel	O
(	O
the	O
"	O
map	O
"	O
step	O
)	O
.	O
</s>
<s>
Therefore	O
,	O
an	O
implementation	O
of	O
the	O
MapReduce	B-Operating_System
framework	O
was	O
adopted	O
by	O
an	O
Apache	O
open-source	O
project	O
named	O
"	O
Hadoop	B-Application
"	O
.	O
</s>
<s>
Apache	B-Language
Spark	I-Language
was	O
developed	O
in	O
2012	O
in	O
response	O
to	O
limitations	O
in	O
the	O
MapReduce	B-Operating_System
paradigm	O
,	O
as	O
it	O
adds	O
in-memory	B-General_Concept
processing	I-General_Concept
and	O
the	O
ability	O
to	O
set	O
up	O
many	O
operations	O
(	O
not	O
just	O
map	O
followed	O
by	O
reducing	O
)	O
.	O
</s>
<s>
MIKE2.0	O
is	O
an	O
open	O
approach	O
to	O
information	O
management	O
that	O
acknowledges	O
the	O
need	O
for	O
revisions	O
due	O
to	O
big	B-Application
data	I-Application
implications	O
identified	O
in	O
an	O
article	O
titled	O
"	O
Big	B-Application
Data	I-Application
Solution	O
Offering	O
"	O
.	O
</s>
<s>
The	O
methodology	O
addresses	O
handling	O
big	B-Application
data	I-Application
in	O
terms	O
of	O
useful	O
permutations	B-Algorithm
of	O
data	O
sources	O
,	O
complexity	O
in	O
interrelationships	O
,	O
and	O
difficulty	O
in	O
deleting	O
(	O
or	O
modifying	O
)	O
individual	O
records	O
.	O
</s>
<s>
Studies	O
in	O
2012	O
showed	O
that	O
a	O
multiple-layer	O
architecture	O
was	O
one	O
option	O
to	O
address	O
the	O
issues	O
that	O
big	B-Application
data	I-Application
presents	O
.	O
</s>
<s>
A	O
distributed	O
parallel	O
architecture	O
distributes	O
data	O
across	O
multiple	O
servers	O
;	O
these	O
parallel	O
execution	O
environments	O
can	O
dramatically	O
improve	O
data	B-General_Concept
processing	I-General_Concept
speeds	O
.	O
</s>
<s>
This	O
type	O
of	O
architecture	O
inserts	O
data	O
into	O
a	O
parallel	O
DBMS	O
,	O
which	O
implements	O
the	O
use	O
of	O
MapReduce	B-Operating_System
and	O
Hadoop	B-Application
frameworks	O
.	O
</s>
<s>
The	O
data	B-General_Concept
lake	I-General_Concept
allows	O
an	O
organization	O
to	O
shift	O
its	O
focus	O
from	O
centralized	O
control	O
to	O
a	O
shared	O
model	O
to	O
respond	O
to	O
the	O
changing	O
dynamics	O
of	O
information	O
management	O
.	O
</s>
<s>
This	O
enables	O
quick	O
segregation	O
of	O
data	O
into	O
the	O
data	B-General_Concept
lake	I-General_Concept
,	O
thereby	O
reducing	O
the	O
overhead	O
time	O
.	O
</s>
<s>
A	O
2011	O
McKinsey	O
Global	O
Institute	O
report	O
characterizes	O
the	O
main	O
components	O
and	O
ecosystem	O
of	O
big	B-Application
data	I-Application
as	O
follows	O
:	O
</s>
<s>
Multidimensional	O
big	B-Application
data	I-Application
can	O
also	O
be	O
represented	O
as	O
OLAP	B-Application
data	O
cubes	O
or	O
,	O
mathematically	O
,	O
tensors	B-Device
.	O
</s>
<s>
Array	B-General_Concept
database	I-General_Concept
systems	I-General_Concept
have	O
set	O
out	O
to	O
provide	O
storage	O
and	O
high-level	O
query	O
support	O
on	O
this	O
data	O
type	O
.	O
</s>
<s>
Additional	O
technologies	O
being	O
applied	O
to	O
big	B-Application
data	I-Application
include	O
efficient	O
tensor-based	O
computation	O
,	O
such	O
as	O
multilinear	O
subspace	O
learning	O
,	O
massively	O
parallel-processing	O
(	O
MPP	B-Operating_System
)	O
databases	O
,	O
search-based	B-Application
applications	I-Application
,	O
data	B-Application
mining	I-Application
,	O
distributed	B-Protocol
file	I-Protocol
systems	I-Protocol
,	O
distributed	O
cache	O
(	O
e.g.	O
,	O
burst	B-Operating_System
buffer	I-Operating_System
and	O
Memcached	B-Application
)	O
,	O
distributed	B-Operating_System
databases	I-Operating_System
,	O
cloud	B-Architecture
and	O
HPC-based	B-Architecture
infrastructure	O
(	O
applications	O
,	O
storage	O
and	O
computing	O
resources	O
)	O
,	O
and	O
the	O
Internet	O
.	O
</s>
<s>
Although	O
,	O
many	O
approaches	O
and	O
technologies	O
have	O
been	O
developed	O
,	O
it	O
still	O
remains	O
difficult	O
to	O
carry	O
out	O
machine	O
learning	O
with	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Some	O
MPP	B-Operating_System
relational	B-Application
databases	I-Application
have	O
the	O
ability	O
to	O
store	O
and	O
manage	O
petabytes	O
of	O
data	O
.	O
</s>
<s>
DARPA	O
's	O
Topological	B-Algorithm
Data	I-Algorithm
Analysis	I-Algorithm
program	O
seeks	O
the	O
fundamental	O
structure	O
of	O
massive	O
data	B-General_Concept
sets	I-General_Concept
and	O
in	O
2008	O
the	O
technology	O
went	O
public	O
with	O
the	O
launch	O
of	O
a	O
company	O
called	O
"	O
Ayasdi	O
"	O
.	O
</s>
<s>
The	O
practitioners	O
of	O
big	B-Application
data	I-Application
analytics	I-Application
processes	O
are	O
generally	O
hostile	O
to	O
slower	O
shared	O
storage	O
,	O
preferring	O
direct-attached	B-General_Concept
storage	I-General_Concept
(	O
DAS	B-General_Concept
)	O
in	O
its	O
various	O
forms	O
from	O
solid	B-Device
state	I-Device
drive	I-Device
(	O
SSD	B-Device
)	O
to	O
high	O
capacity	O
SATA	O
disk	O
buried	O
inside	O
parallel	O
processing	O
nodes	O
.	O
</s>
<s>
The	O
perception	O
of	O
shared	O
storage	O
architectures	O
—	O
storage	O
area	O
network	O
(	O
SAN	O
)	O
and	O
network-attached	B-Application
storage	I-Application
(	O
NAS	B-Application
)	O
—	O
is	O
that	O
they	O
are	O
relatively	O
slow	O
,	O
complex	O
,	O
and	O
expensive	O
.	O
</s>
<s>
These	O
qualities	O
are	O
not	O
consistent	O
with	O
big	B-Application
data	I-Application
analytics	I-Application
systems	O
that	O
thrive	O
on	O
system	O
performance	O
,	O
commodity	O
infrastructure	O
,	O
and	O
low	O
cost	O
.	O
</s>
<s>
Real	O
or	O
near-real-time	O
information	O
delivery	O
is	O
one	O
of	O
the	O
defining	O
characteristics	O
of	O
big	B-Application
data	I-Application
analytics	I-Application
.	O
</s>
<s>
Data	O
in	O
direct-attached	O
memory	O
or	O
disk	O
is	O
good	O
—	O
data	O
on	O
memory	O
or	O
disk	O
at	O
the	O
other	O
end	O
of	O
an	O
FC	B-Device
SAN	O
connection	O
is	O
not	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
has	O
increased	O
the	O
demand	O
of	O
information	O
management	O
specialists	O
so	O
much	O
so	O
that	O
Software	O
AG	O
,	O
Oracle	B-Application
Corporation	I-Application
,	O
IBM	O
,	O
Microsoft	O
,	O
SAP	B-Application
,	O
EMC	O
,	O
HP	O
,	O
and	O
Dell	O
have	O
spent	O
more	O
than	O
$15billion	O
on	O
software	O
firms	O
specializing	O
in	O
data	O
management	O
and	O
analytics	O
.	O
</s>
<s>
According	O
to	O
one	O
estimate	O
,	O
one-third	O
of	O
the	O
globally	O
stored	O
information	O
is	O
in	O
the	O
form	O
of	O
alphanumeric	O
text	O
and	O
still	O
image	O
data	O
,	O
which	O
is	O
the	O
format	O
most	O
useful	O
for	O
most	O
big	B-Application
data	I-Application
applications	O
.	O
</s>
<s>
While	O
many	O
vendors	O
offer	O
off-the-shelf	O
products	O
for	O
big	B-Application
data	I-Application
,	O
experts	O
promote	O
the	O
development	O
of	O
in-house	O
custom-tailored	O
systems	O
if	O
the	O
company	O
has	O
sufficient	O
technical	O
capabilities	O
.	O
</s>
<s>
The	O
use	O
and	O
adoption	O
of	O
big	B-Application
data	I-Application
within	O
governmental	O
processes	O
allows	O
efficiencies	O
in	O
terms	O
of	O
cost	O
,	O
productivity	O
,	O
and	O
innovation	O
,	O
but	O
does	O
not	O
come	O
without	O
its	O
flaws	O
.	O
</s>
<s>
Data	B-General_Concept
analysis	I-General_Concept
often	O
requires	O
multiple	O
parts	O
of	O
government	O
(	O
central	O
and	O
local	O
)	O
to	O
work	O
in	O
collaboration	O
and	O
create	O
new	O
and	O
innovative	O
processes	O
to	O
deliver	O
the	O
desired	O
outcome	O
.	O
</s>
<s>
A	O
common	O
government	O
organization	O
that	O
makes	O
use	O
of	O
big	B-Application
data	I-Application
is	O
the	O
National	O
Security	O
Administration	O
(	O
NSA	O
)	O
,	O
which	O
monitors	O
the	O
activities	O
of	O
the	O
Internet	O
constantly	O
in	O
search	O
for	O
potential	O
patterns	O
of	O
suspicious	O
or	O
illegal	O
activities	O
their	O
system	O
may	O
pick	O
up	O
.	O
</s>
<s>
CRVS	O
is	O
a	O
source	O
of	O
big	B-Application
data	I-Application
for	O
governments	O
.	O
</s>
<s>
Research	O
on	O
the	O
effective	O
usage	O
of	O
information	O
and	O
communication	O
technologies	O
for	O
development	O
(	O
also	O
known	O
as	O
"	O
ICT4D	O
"	O
)	O
suggests	O
that	O
big	B-Application
data	I-Application
technology	O
can	O
make	O
important	O
contributions	O
but	O
also	O
present	O
unique	O
challenges	O
to	O
international	O
development	O
.	O
</s>
<s>
Advancements	O
in	O
big	B-Application
data	I-Application
analysis	I-Application
offer	O
cost-effective	O
opportunities	O
to	O
improve	O
decision-making	O
in	O
critical	O
development	O
areas	O
such	O
as	O
health	O
care	O
,	O
employment	O
,	O
economic	O
productivity	O
,	O
crime	O
,	O
security	O
,	O
and	O
natural	O
disaster	O
and	O
resource	O
management	O
.	O
</s>
<s>
However	O
,	O
longstanding	O
challenges	O
for	O
developing	O
regions	O
such	O
as	O
inadequate	O
technological	O
infrastructure	O
and	O
economic	O
and	O
human	O
resource	O
scarcity	O
exacerbate	O
existing	O
concerns	O
with	O
big	B-Application
data	I-Application
such	O
as	O
privacy	O
,	O
imperfect	O
methodology	O
,	O
and	O
interoperability	O
issues	O
.	O
</s>
<s>
The	O
challenge	O
of	O
"	O
big	B-Application
data	I-Application
for	O
development	O
"	O
is	O
currently	O
evolving	O
toward	O
the	O
application	O
of	O
this	O
data	O
through	O
machine	O
learning	O
,	O
known	O
as	O
"	O
artificial	B-Application
intelligence	I-Application
for	I-Application
development	I-Application
(	O
AI4D	O
)	O
.	O
</s>
<s>
A	O
major	O
practical	O
application	O
of	O
big	B-Application
data	I-Application
for	O
development	O
has	O
been	O
"	O
fighting	O
poverty	O
with	O
data	O
"	O
.	O
</s>
<s>
Big	B-Application
Data	I-Application
is	O
being	O
rapidly	O
adopted	O
in	O
Finance	O
to	O
1	O
)	O
speed	O
up	O
processing	O
and	O
2	O
)	O
deliver	O
better	O
,	O
more	O
informed	O
inferences	O
,	O
both	O
internally	O
and	O
to	O
the	O
clients	O
of	O
the	O
financial	O
institutions	O
..	O
</s>
<s>
The	O
financial	O
applications	O
of	O
Big	B-Application
Data	I-Application
range	O
from	O
investing	O
decisions	O
and	O
trading	O
(	O
processing	O
volumes	O
of	O
available	O
price	O
data	O
,	O
limit	O
order	O
books	O
,	O
economic	O
data	O
and	O
more	O
,	O
all	O
at	O
the	O
same	O
time	O
)	O
,	O
portfolio	O
management	O
(	O
optimizing	O
over	O
an	O
increasingly	O
large	O
array	O
of	O
financial	O
instruments	O
,	O
potentially	O
selected	O
from	O
different	O
asset	O
classes	O
)	O
,	O
risk	O
management	O
(	O
credit	O
rating	O
based	O
on	O
extended	O
information	O
)	O
,	O
and	O
any	O
other	O
aspect	O
where	O
the	O
data	O
inputs	O
are	O
large	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
analytics	I-Application
was	O
used	O
in	O
healthcare	O
by	O
providing	O
personalized	O
medicine	O
and	O
prescriptive	O
analytics	O
,	O
clinical	O
risk	O
intervention	O
and	O
predictive	B-General_Concept
analytics	I-General_Concept
,	O
waste	O
and	O
care	O
variability	O
reduction	O
,	O
automated	O
external	O
and	O
internal	O
reporting	O
of	O
patient	O
data	O
,	O
standardized	O
medical	O
terms	O
and	O
patient	O
registries	O
.	O
</s>
<s>
This	O
includes	O
electronic	B-Application
health	I-Application
record	I-Application
data	O
,	O
imaging	O
data	O
,	O
patient	O
generated	O
data	O
,	O
sensor	O
data	O
,	O
and	O
other	O
forms	O
of	O
difficult	O
to	O
process	O
data	O
.	O
</s>
<s>
"	O
Big	B-Application
data	I-Application
very	O
often	O
means	O
'	O
dirty	O
data	O
 '	O
and	O
the	O
fraction	O
of	O
data	O
inaccuracies	O
increases	O
with	O
data	O
volume	O
growth.	O
"	O
</s>
<s>
Human	O
inspection	O
at	O
the	O
big	B-Application
data	I-Application
scale	O
is	O
impossible	O
and	O
there	O
is	O
a	O
desperate	O
need	O
in	O
health	O
service	O
for	O
intelligent	O
tools	O
for	O
accuracy	O
and	O
believability	O
control	O
and	O
handling	O
of	O
information	O
missed	O
.	O
</s>
<s>
While	O
extensive	O
information	O
in	O
healthcare	O
is	O
now	O
electronic	O
,	O
it	O
fits	O
under	O
the	O
big	B-Application
data	I-Application
umbrella	O
as	O
most	O
is	O
unstructured	O
and	O
difficult	O
to	O
use	O
.	O
</s>
<s>
The	O
use	O
of	O
big	B-Application
data	I-Application
in	O
healthcare	O
has	O
raised	O
significant	O
ethical	O
challenges	O
ranging	O
from	O
risks	O
for	O
individual	O
rights	O
,	O
privacy	O
and	O
autonomy	O
,	O
to	O
transparency	O
and	O
trust	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
in	O
health	O
research	O
is	O
particularly	O
promising	O
in	O
terms	O
of	O
exploratory	O
biomedical	O
research	O
,	O
as	O
data-driven	O
analysis	O
can	O
move	O
forward	O
more	O
quickly	O
than	O
hypothesis-driven	O
research	O
.	O
</s>
<s>
Then	O
,	O
trends	O
seen	O
in	O
data	B-General_Concept
analysis	I-General_Concept
can	O
be	O
tested	O
in	O
traditional	O
,	O
hypothesis-driven	O
follow	O
up	O
biological	O
research	O
and	O
eventually	O
clinical	O
research	O
.	O
</s>
<s>
A	O
related	O
application	O
sub-area	O
,	O
that	O
heavily	O
relies	O
on	O
big	B-Application
data	I-Application
,	O
within	O
the	O
healthcare	O
field	O
is	O
that	O
of	O
computer-aided	O
diagnosis	O
in	O
medicine	O
.	O
</s>
<s>
These	O
are	O
just	O
a	O
few	O
of	O
the	O
many	O
examples	O
where	O
computer-aided	O
diagnosis	O
uses	O
big	B-Application
data	I-Application
.	O
</s>
<s>
For	O
this	O
reason	O
,	O
big	B-Application
data	I-Application
has	O
been	O
recognized	O
as	O
one	O
of	O
the	O
seven	O
key	O
challenges	O
that	O
computer-aided	O
diagnosis	O
systems	O
need	O
to	O
overcome	O
in	O
order	O
to	O
reach	O
the	O
next	O
level	O
of	O
performance	O
.	O
</s>
<s>
To	O
understand	O
how	O
the	O
media	O
uses	O
big	B-Application
data	I-Application
,	O
it	O
is	O
first	O
necessary	O
to	O
provide	O
some	O
context	O
into	O
the	O
mechanism	O
used	O
for	O
media	O
process	O
.	O
</s>
<s>
It	O
has	O
been	O
suggested	O
by	O
Nick	O
Couldry	O
and	O
Joseph	O
Turow	O
that	O
practitioners	O
in	O
media	O
and	O
advertising	O
approach	O
big	B-Application
data	I-Application
as	O
many	O
actionable	O
points	O
of	O
information	O
about	O
millions	O
of	O
individuals	O
.	O
</s>
<s>
For	O
example	O
,	O
publishing	O
environments	O
are	O
increasingly	O
tailoring	O
messages	O
(	O
advertisements	O
)	O
and	O
content	O
(	O
articles	O
)	O
to	O
appeal	O
to	O
consumers	O
that	O
have	O
been	O
exclusively	O
gleaned	O
through	O
various	O
data-mining	B-Application
activities	O
.	O
</s>
<s>
Data	B-General_Concept
journalism	I-General_Concept
:	O
publishers	O
and	O
journalists	O
use	O
big	B-Application
data	I-Application
tools	O
to	O
provide	O
unique	O
and	O
innovative	O
insights	O
and	O
infographics	O
.	O
</s>
<s>
Channel	O
4	O
,	O
the	O
British	O
public-service	O
television	O
broadcaster	O
,	O
is	O
a	O
leader	O
in	O
the	O
field	O
of	O
big	B-Application
data	I-Application
and	O
data	B-General_Concept
analysis	I-General_Concept
.	O
</s>
<s>
Big	B-Application
data	I-Application
and	O
the	O
IoT	B-Operating_System
work	O
in	O
conjunction	O
.	O
</s>
<s>
Data	O
extracted	O
from	O
IoT	B-Operating_System
devices	O
provides	O
a	O
mapping	O
of	O
device	O
inter-connectivity	O
.	O
</s>
<s>
The	O
IoT	B-Operating_System
is	O
also	O
increasingly	O
adopted	O
as	O
a	O
means	O
of	O
gathering	O
sensory	O
data	O
,	O
and	O
this	O
sensory	O
data	O
has	O
been	O
used	O
in	O
medical	O
,	O
manufacturing	O
and	O
transportation	O
contexts	O
.	O
</s>
<s>
Kevin	O
Ashton	O
,	O
the	O
digital	O
innovation	O
expert	O
who	O
is	O
credited	O
with	O
coining	O
the	O
term	O
,	O
defines	O
the	B-Operating_System
Internet	I-Operating_System
of	I-Operating_System
things	I-Operating_System
in	O
this	O
quote	O
:	O
"	O
If	O
we	O
had	O
computers	O
that	O
knew	O
everything	O
there	O
was	O
to	O
know	O
about	O
things	O
—	O
using	O
data	O
they	O
gathered	O
without	O
any	O
help	O
from	O
us	O
—	O
we	O
would	O
be	O
able	O
to	O
track	O
and	O
count	O
everything	O
,	O
and	O
greatly	O
reduce	O
waste	O
,	O
loss	O
,	O
and	O
cost	O
.	O
</s>
<s>
Especially	O
since	O
2015	O
,	O
big	B-Application
data	I-Application
has	O
come	O
to	O
prominence	O
within	O
business	O
operations	O
as	O
a	O
tool	O
to	O
help	O
employees	O
work	O
more	O
efficiently	O
and	O
streamline	O
the	O
collection	O
and	O
distribution	O
of	O
information	B-General_Concept
technology	I-General_Concept
(	O
IT	O
)	O
.	O
</s>
<s>
The	O
use	O
of	O
big	B-Application
data	I-Application
to	O
resolve	O
IT	O
and	O
data	O
collection	O
issues	O
within	O
an	O
enterprise	O
is	O
called	O
IT	B-Architecture
operations	I-Architecture
analytics	I-Architecture
(	O
ITOA	B-Architecture
)	O
.	O
</s>
<s>
By	O
applying	O
big	B-Application
data	I-Application
principles	O
into	O
the	O
concepts	O
of	O
machine	B-Application
intelligence	I-Application
and	O
deep	O
computing	O
,	O
IT	O
departments	O
can	O
predict	O
potential	O
issues	O
and	O
prevent	O
them	O
.	O
</s>
<s>
ITOA	B-Architecture
businesses	O
offer	O
platforms	O
for	O
systems	B-Application
management	I-Application
that	O
bring	O
data	B-Application
silos	I-Application
together	O
and	O
generate	O
insights	O
from	O
the	O
whole	O
of	O
the	O
system	O
rather	O
than	O
from	O
isolated	O
pockets	O
of	O
data	O
.	O
</s>
<s>
The	O
Social	B-Application
Credit	I-Application
System	I-Application
,	O
now	O
being	O
piloted	O
in	O
a	O
number	O
of	O
Chinese	O
cities	O
,	O
is	O
considered	O
a	O
form	O
of	O
mass	O
surveillance	O
which	O
uses	O
big	B-Application
data	I-Application
analysis	I-Application
technology	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
analysis	I-Application
was	O
tried	O
out	O
for	O
the	O
BJP	O
to	O
win	O
the	O
2014	O
Indian	O
General	O
Election	O
.	O
</s>
<s>
Personalized	O
diabetic	O
treatments	O
can	O
be	O
created	O
through	O
GlucoMe	O
's	O
big	B-Application
data	I-Application
solution	O
.	O
</s>
<s>
Examples	O
of	O
uses	O
of	O
big	B-Application
data	I-Application
in	O
public	O
services	O
:	O
</s>
<s>
In	O
2012	O
,	O
the	O
Obama	O
administration	O
announced	O
the	O
Big	B-Application
Data	I-Application
Research	O
and	O
Development	O
Initiative	O
,	O
to	O
explore	O
how	O
big	B-Application
data	I-Application
could	O
be	O
used	O
to	O
address	O
important	O
problems	O
faced	O
by	O
the	O
government	O
.	O
</s>
<s>
The	O
initiative	O
is	O
composed	O
of	O
84	O
different	O
big	B-Application
data	I-Application
programs	O
spread	O
across	O
six	O
departments	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
analysis	I-Application
played	O
a	O
large	O
role	O
in	O
Barack	O
Obama	O
's	O
successful	O
2012	O
re-election	O
campaign	O
.	O
</s>
<s>
The	O
United	O
States	O
Federal	O
Government	O
owns	O
five	O
of	O
the	O
ten	O
most	O
powerful	O
supercomputers	B-Architecture
in	O
the	O
world	O
.	O
</s>
<s>
After	O
filtering	B-Application
and	O
refraining	O
from	O
recording	O
more	O
than	O
99.99995	O
%	O
of	O
these	O
streams	O
,	O
there	O
are	O
1,000	O
collisions	O
of	O
interest	O
per	O
second	O
.	O
</s>
<s>
The	O
NASA	O
Center	O
for	O
Climate	O
Simulation	O
(	O
NCCS	O
)	O
stores	O
32	O
petabytes	O
of	O
climate	O
observations	O
and	O
simulations	O
on	O
the	O
Discover	O
supercomputing	B-Architecture
cluster	O
.	O
</s>
<s>
Google	B-Application
's	I-Application
DNAStack	O
compiles	O
and	O
organizes	O
DNA	O
samples	O
of	O
genetic	O
data	O
from	O
around	O
the	O
world	O
to	O
identify	O
diseases	O
and	O
other	O
medical	O
defects	O
.	O
</s>
<s>
DNAStack	O
,	O
a	O
part	O
of	O
Google	B-Application
Genomics	O
,	O
allows	O
scientists	O
to	O
use	O
the	O
vast	O
sample	O
of	O
resources	O
from	O
Google	B-Application
's	I-Application
search	O
server	O
to	O
scale	O
social	O
experiments	O
that	O
would	O
usually	O
take	O
years	O
,	O
instantly	O
.	O
</s>
<s>
Computational	O
fluid	O
dynamics	O
(	O
CFD	O
)	O
and	O
hydrodynamic	O
turbulence	O
research	O
generate	O
massive	O
data	B-General_Concept
sets	I-General_Concept
.	O
</s>
<s>
Big	B-Application
data	I-Application
can	O
be	O
used	O
to	O
improve	O
training	O
and	O
understanding	O
competitors	O
,	O
using	O
sport	O
sensors	O
.	O
</s>
<s>
It	O
is	O
also	O
possible	O
to	O
predict	O
winners	O
in	O
a	O
match	O
using	O
big	B-Application
data	I-Application
analytics	I-Application
.	O
</s>
<s>
Based	O
on	O
the	O
data	O
,	O
engineers	O
and	O
data	B-General_Concept
analysts	I-General_Concept
decide	O
whether	O
adjustments	O
should	O
be	O
made	O
in	O
order	O
to	O
win	O
a	O
race	O
.	O
</s>
<s>
Besides	O
,	O
using	O
big	B-Application
data	I-Application
,	O
race	O
teams	O
try	O
to	O
predict	O
the	O
time	O
they	O
will	O
finish	O
the	O
race	O
beforehand	O
,	O
based	O
on	O
simulations	O
using	O
data	O
collected	O
over	O
the	O
season	O
.	O
</s>
<s>
eBay.com	B-Application
uses	O
two	O
data	O
warehouses	O
at	O
7.5	O
petabytes	O
and	O
40PB	O
as	O
well	O
as	O
a	O
40PB	O
Hadoop	B-Application
cluster	O
for	O
search	O
,	O
consumer	O
recommendations	O
,	O
and	O
merchandising	O
.	O
</s>
<s>
Amazon.com	B-Application
handles	O
millions	O
of	O
back-end	O
operations	O
every	O
day	O
,	O
as	O
well	O
as	O
queries	O
from	O
more	O
than	O
half	O
a	O
million	O
third-party	O
sellers	O
.	O
</s>
<s>
The	O
core	O
technology	O
that	O
keeps	O
Amazon	B-Application
running	O
is	O
Linux-based	O
and	O
they	O
had	O
the	O
world	O
's	O
three	O
largest	O
Linux	O
databases	O
,	O
with	O
capacities	O
of	O
7.8	O
TB	O
,	O
18.5	O
TB	O
,	O
and	O
24.7	O
TB	O
.	O
</s>
<s>
Facebook	B-Application
handles	O
50billion	O
photos	O
from	O
its	O
user	O
base	O
.	O
</s>
<s>
,	O
Facebook	B-Application
reached	O
2	O
billion	O
monthly	O
active	O
users	O
.	O
</s>
<s>
Google	B-Application
was	O
handling	O
roughly	O
100billion	O
searches	O
per	O
month	O
.	O
</s>
<s>
During	O
the	O
COVID-19	O
pandemic	O
,	O
big	B-Application
data	I-Application
was	O
raised	O
as	O
a	O
way	O
to	O
minimise	O
the	O
impact	O
of	O
the	O
disease	O
.	O
</s>
<s>
Significant	O
applications	O
of	O
big	B-Application
data	I-Application
included	O
minimising	O
the	O
spread	O
of	O
the	O
virus	O
,	O
case	O
identification	O
and	O
development	O
of	O
medical	O
treatment	O
.	O
</s>
<s>
Governments	O
used	O
big	B-Application
data	I-Application
to	O
track	O
infected	O
people	O
to	O
minimise	O
spread	O
.	O
</s>
<s>
Encrypted	O
search	O
and	O
cluster	O
formation	O
in	O
big	B-Application
data	I-Application
were	O
demonstrated	O
in	O
March	O
2014	O
at	O
the	O
American	O
Society	O
of	O
Engineering	O
Education	O
.	O
</s>
<s>
Gautam	O
Siwach	O
engaged	O
at	O
Tackling	O
the	O
challenges	O
of	O
Big	B-Application
Data	I-Application
by	O
MIT	O
Computer	O
Science	O
and	O
Artificial	B-Application
Intelligence	I-Application
Laboratory	O
and	O
Amir	O
Esmailpour	O
at	O
the	O
UNH	O
Research	O
Group	O
investigated	O
the	O
key	O
features	O
of	O
big	B-Application
data	I-Application
as	O
the	O
formation	O
of	O
clusters	O
and	O
their	O
interconnections	O
.	O
</s>
<s>
They	O
focused	O
on	O
the	O
security	O
of	O
big	B-Application
data	I-Application
and	O
the	O
orientation	O
of	O
the	O
term	O
towards	O
the	O
presence	O
of	O
different	O
types	O
of	O
data	O
in	O
an	O
encrypted	O
form	O
at	O
cloud	B-Architecture
interface	O
by	O
providing	O
the	O
raw	O
definitions	O
and	O
real-time	O
examples	O
within	O
the	O
technology	O
.	O
</s>
<s>
Moreover	O
,	O
they	O
proposed	O
an	O
approach	O
for	O
identifying	O
the	O
encoding	O
technique	O
to	O
advance	O
towards	O
an	O
expedited	O
search	O
over	O
encrypted	O
text	O
leading	O
to	O
the	O
security	O
enhancements	O
in	O
big	B-Application
data	I-Application
.	O
</s>
<s>
In	O
March	O
2012	O
,	O
The	O
White	O
House	O
announced	O
a	O
national	O
"	O
Big	B-Application
Data	I-Application
Initiative	O
"	O
that	O
consisted	O
of	O
six	O
federal	O
departments	O
and	O
agencies	O
committing	O
more	O
than	O
$200million	O
to	O
big	B-Application
data	I-Application
research	O
projects	O
.	O
</s>
<s>
The	O
AMPLab	O
also	O
received	O
funds	O
from	O
DARPA	O
,	O
and	O
over	O
a	O
dozen	O
industrial	O
sponsors	O
and	O
uses	O
big	B-Application
data	I-Application
to	O
attack	O
a	O
wide	O
range	O
of	O
problems	O
from	O
predicting	O
traffic	O
congestion	O
to	O
fighting	O
cancer	O
.	O
</s>
<s>
The	O
White	O
House	O
Big	B-Application
Data	I-Application
Initiative	O
also	O
included	O
a	O
commitment	O
by	O
the	O
Department	O
of	O
Energy	O
to	O
provide	O
$25	O
million	O
in	O
funding	O
over	O
five	O
years	O
to	O
establish	O
the	O
Scalable	O
Data	O
Management	O
,	O
Analysis	O
and	O
Visualization	B-Application
(	O
SDAV	O
)	O
Institute	O
,	O
led	O
by	O
the	O
Energy	O
Department	O
's	O
Lawrence	O
Berkeley	O
National	O
Laboratory	O
.	O
</s>
<s>
The	O
SDAV	O
Institute	O
aims	O
to	O
bring	O
together	O
the	O
expertise	O
of	O
six	O
national	O
laboratories	O
and	O
seven	O
universities	O
to	O
develop	O
new	O
tools	O
to	O
help	O
scientists	O
manage	O
and	O
visualize	O
data	O
on	O
the	O
department	O
's	O
supercomputers	B-Architecture
.	O
</s>
<s>
The	O
U.S.	O
state	O
of	O
Massachusetts	O
announced	O
the	O
Massachusetts	O
Big	B-Application
Data	I-Application
Initiative	O
in	O
May	O
2012	O
,	O
which	O
provides	O
funding	O
from	O
the	O
state	O
government	O
and	O
private	O
companies	O
to	O
a	O
variety	O
of	O
research	O
institutions	O
.	O
</s>
<s>
The	O
Massachusetts	O
Institute	O
of	O
Technology	O
hosts	O
the	O
Intel	O
Science	O
and	O
Technology	O
Center	O
for	O
Big	B-Application
Data	I-Application
in	O
the	O
MIT	O
Computer	O
Science	O
and	O
Artificial	B-Application
Intelligence	I-Application
Laboratory	O
,	O
combining	O
government	O
,	O
corporate	O
,	O
and	O
institutional	O
funding	O
and	O
research	O
efforts	O
.	O
</s>
<s>
The	O
European	O
Commission	O
is	O
funding	O
the	O
two-year-long	O
Big	B-Application
Data	I-Application
Public	O
Private	O
Forum	O
through	O
their	O
Seventh	O
Framework	O
Program	O
to	O
engage	O
companies	O
,	O
academics	O
and	O
other	O
stakeholders	O
in	O
discussing	O
big	B-Application
data	I-Application
issues	O
.	O
</s>
<s>
The	O
project	O
aims	O
to	O
define	O
a	O
strategy	O
in	O
terms	O
of	O
research	O
and	O
innovation	O
to	O
guide	O
supporting	O
actions	O
from	O
the	O
European	O
Commission	O
in	O
the	O
successful	O
implementation	O
of	O
the	O
big	B-Application
data	I-Application
economy	O
.	O
</s>
<s>
The	O
British	O
government	O
announced	O
in	O
March	O
2014	O
the	O
founding	O
of	O
the	O
Alan	O
Turing	O
Institute	O
,	O
named	O
after	O
the	O
computer	O
pioneer	O
and	O
code-breaker	O
,	O
which	O
will	O
focus	O
on	O
new	O
ways	O
to	O
collect	O
and	O
analyze	O
large	O
data	B-General_Concept
sets	I-General_Concept
.	O
</s>
<s>
At	O
the	O
University	O
of	O
Waterloo	O
Stratford	O
Campus	O
Canadian	O
Open	O
Data	O
Experience	O
(	O
CODE	O
)	O
Inspiration	O
Day	O
,	O
participants	O
demonstrated	O
how	O
using	O
data	B-Application
visualization	I-Application
can	O
increase	O
the	O
understanding	O
and	O
appeal	O
of	O
big	B-Application
data	I-Application
sets	O
and	O
communicate	O
their	O
story	O
to	O
the	O
world	O
.	O
</s>
<s>
Computational	O
social	O
sciences	O
–	O
Anyone	O
can	O
use	O
application	O
programming	O
interfaces	O
(	O
APIs	O
)	O
provided	O
by	O
big	B-Application
data	I-Application
holders	O
,	O
such	O
as	O
Google	B-Application
and	O
Twitter	O
,	O
to	O
do	O
research	O
in	O
the	O
social	O
and	O
behavioral	O
sciences	O
.	O
</s>
<s>
used	O
Google	B-Application
Trends	I-Application
data	O
to	O
demonstrate	O
that	O
Internet	O
users	O
from	O
countries	O
with	O
a	O
higher	O
per	O
capita	O
gross	O
domestic	O
products	O
(	O
GDPs	O
)	O
are	O
more	O
likely	O
to	O
search	O
for	O
information	O
about	O
the	O
future	O
than	O
information	O
about	O
the	O
past	O
.	O
</s>
<s>
The	O
authors	O
of	O
the	O
study	O
examined	O
Google	B-Application
queries	O
logs	O
made	O
by	O
ratio	O
of	O
the	O
volume	O
of	O
searches	O
for	O
the	O
coming	O
year	O
(	O
2011	O
)	O
to	O
the	O
volume	O
of	O
searches	O
for	O
the	O
previous	O
year	O
(	O
2009	O
)	O
,	O
which	O
they	O
call	O
the	O
"	O
future	B-Application
orientation	I-Application
index	I-Application
"	O
.	O
</s>
<s>
They	O
compared	O
the	O
future	B-Application
orientation	I-Application
index	I-Application
to	O
the	O
per	O
capita	O
GDP	O
of	O
each	O
country	O
,	O
and	O
found	O
a	O
strong	O
tendency	O
for	O
countries	O
where	O
Google	B-Application
users	O
inquire	O
more	O
about	O
the	O
future	O
to	O
have	O
a	O
higher	O
GDP	O
.	O
</s>
<s>
Tobias	O
Preis	O
and	O
his	O
colleagues	O
Helen	O
Susannah	O
Moat	O
and	O
H	O
.	O
Eugene	O
Stanley	O
introduced	O
a	O
method	O
to	O
identify	O
online	O
precursors	O
for	O
stock	O
market	O
moves	O
,	O
using	O
trading	O
strategies	O
based	O
on	O
search	O
volume	O
data	O
provided	O
by	O
Google	B-Application
Trends	I-Application
.	O
</s>
<s>
Their	O
analysis	O
of	O
Google	B-Application
search	O
volume	O
for	O
98	O
terms	O
of	O
varying	O
financial	O
relevance	O
,	O
published	O
in	O
Scientific	O
Reports	O
,	O
suggests	O
that	O
increases	O
in	O
search	O
volume	O
for	O
financially	O
relevant	O
search	O
terms	O
tend	O
to	O
precede	O
large	O
losses	O
in	O
financial	O
markets	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
sets	O
come	O
with	O
algorithmic	O
challenges	O
that	O
previously	O
did	O
not	O
exist	O
.	O
</s>
<s>
The	O
Workshops	O
on	O
Algorithms	O
for	O
Modern	O
Massive	O
Data	B-General_Concept
Sets	I-General_Concept
(	O
MMDS	O
)	O
bring	O
together	O
computer	O
scientists	O
,	O
statisticians	O
,	O
mathematicians	O
,	O
and	O
data	B-General_Concept
analysis	I-General_Concept
practitioners	O
to	O
discuss	O
algorithmic	O
challenges	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Regarding	O
big	B-Application
data	I-Application
,	O
such	O
concepts	O
of	O
magnitude	O
are	O
relative	O
.	O
</s>
<s>
As	O
it	O
is	O
stated	O
"	O
If	O
the	O
past	O
is	O
of	O
any	O
guidance	O
,	O
then	O
today	O
's	O
big	B-Application
data	I-Application
most	O
likely	O
will	O
not	O
be	O
considered	O
as	O
such	O
in	O
the	O
near	O
future.	O
"	O
</s>
<s>
A	O
research	O
question	O
that	O
is	O
asked	O
about	O
big	B-Application
data	I-Application
sets	O
is	O
whether	O
it	O
is	O
necessary	O
to	O
look	O
at	O
the	O
full	O
data	O
to	O
draw	O
certain	O
conclusions	O
about	O
the	O
properties	O
of	O
the	O
data	O
or	O
if	O
is	O
a	O
sample	O
is	O
good	O
enough	O
.	O
</s>
<s>
The	O
name	O
big	B-Application
data	I-Application
itself	O
contains	O
a	O
term	O
related	O
to	O
size	O
and	O
this	O
is	O
an	O
important	O
characteristic	O
of	O
big	B-Application
data	I-Application
.	O
</s>
<s>
But	O
sampling	O
enables	O
the	O
selection	O
of	O
right	O
data	O
points	O
from	O
within	O
the	O
larger	O
data	B-General_Concept
set	I-General_Concept
to	O
estimate	O
the	O
characteristics	O
of	O
the	O
whole	O
population	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
can	O
be	O
broken	O
down	O
by	O
various	O
data	O
point	O
categories	O
such	O
as	O
demographic	O
,	O
psychographic	O
,	O
behavioral	O
,	O
and	O
transactional	O
data	O
.	O
</s>
<s>
There	O
has	O
been	O
some	O
work	O
done	O
in	O
sampling	O
algorithms	O
for	O
big	B-Application
data	I-Application
.	O
</s>
<s>
Critiques	O
of	O
the	O
big	B-Application
data	I-Application
paradigm	O
come	O
in	O
two	O
flavors	O
:	O
those	O
that	O
question	O
the	O
implications	O
of	O
the	O
approach	O
itself	O
,	O
and	O
those	O
that	O
question	O
the	O
way	O
it	O
is	O
currently	O
done	O
.	O
</s>
<s>
Mark	O
Graham	O
has	O
leveled	O
broad	O
critiques	O
at	O
Chris	O
Anderson	O
's	O
assertion	O
that	O
big	B-Application
data	I-Application
will	O
spell	O
the	O
end	O
of	O
theory	O
:	O
focusing	O
in	O
particular	O
on	O
the	O
notion	O
that	O
big	B-Application
data	I-Application
must	O
always	O
be	O
contextualized	O
in	O
their	O
social	O
,	O
economic	O
,	O
and	O
political	O
contexts	O
.	O
</s>
<s>
To	O
overcome	O
this	O
insight	O
deficit	O
,	O
big	B-Application
data	I-Application
,	O
no	O
matter	O
how	O
comprehensive	O
or	O
well	O
analyzed	O
,	O
must	O
be	O
complemented	O
by	O
"	O
big	O
judgment	O
"	O
,	O
according	O
to	O
an	O
article	O
in	O
the	O
Harvard	O
Business	O
Review	O
.	O
</s>
<s>
Much	O
in	O
the	O
same	O
line	O
,	O
it	O
has	O
been	O
pointed	O
out	O
that	O
the	O
decisions	O
based	O
on	O
the	O
analysis	O
of	O
big	B-Application
data	I-Application
are	O
inevitably	O
"	O
informed	O
by	O
the	O
world	O
as	O
it	O
was	O
in	O
the	O
past	O
,	O
or	O
,	O
at	O
best	O
,	O
as	O
it	O
currently	O
is	O
"	O
.	O
</s>
<s>
If	O
the	O
system	O
's	O
dynamics	O
of	O
the	O
future	O
change	O
(	O
if	O
it	O
is	O
not	O
a	O
stationary	B-Algorithm
process	I-Algorithm
)	O
,	O
the	O
past	O
can	O
say	O
little	O
about	O
the	O
future	O
.	O
</s>
<s>
Additionally	O
,	O
it	O
has	O
been	O
suggested	O
to	O
combine	O
big	B-Application
data	I-Application
approaches	O
with	O
computer	O
simulations	O
,	O
such	O
as	O
agent-based	O
models	O
and	O
complex	O
systems	O
.	O
</s>
<s>
Finally	O
,	O
the	O
use	O
of	O
multivariate	O
methods	O
that	O
probe	O
for	O
the	O
latent	O
structure	O
of	O
the	O
data	O
,	O
such	O
as	O
factor	O
analysis	O
and	O
cluster	B-Algorithm
analysis	I-Algorithm
,	O
have	O
proven	O
useful	O
as	O
analytic	O
approaches	O
that	O
go	O
well	O
beyond	O
the	O
bi-variate	O
approaches	O
(	O
e.g.	O
</s>
<s>
contingency	B-Application
tables	I-Application
)	O
typically	O
employed	O
with	O
smaller	O
data	B-General_Concept
sets	I-General_Concept
.	O
</s>
<s>
Privacy	O
advocates	O
are	O
concerned	O
about	O
the	O
threat	O
to	O
privacy	O
represented	O
by	O
increasing	O
storage	O
and	O
integration	B-General_Concept
of	O
personally	O
identifiable	O
information	O
;	O
expert	O
panels	O
have	O
released	O
various	O
policy	O
recommendations	O
to	O
conform	O
practice	O
to	O
expectations	O
of	O
privacy	O
.	O
</s>
<s>
The	O
misuse	O
of	O
big	B-Application
data	I-Application
in	O
several	O
cases	O
by	O
media	O
,	O
companies	O
,	O
and	O
even	O
the	O
government	O
has	O
allowed	O
for	O
abolition	O
of	O
trust	O
in	O
almost	O
every	O
fundamental	O
institution	O
holding	O
up	O
society	O
.	O
</s>
<s>
Nayef	O
Al-Rodhan	O
argues	O
that	O
a	O
new	O
kind	O
of	O
social	O
contract	O
will	O
be	O
needed	O
to	O
protect	O
individual	O
liberties	O
in	O
the	O
context	O
of	O
big	B-Application
data	I-Application
and	O
giant	O
corporations	O
that	O
own	O
vast	O
amounts	O
of	O
information	O
,	O
and	O
that	O
the	O
use	O
of	O
big	B-Application
data	I-Application
should	O
be	O
monitored	O
and	O
better	O
regulated	O
at	O
the	O
national	O
and	O
international	O
levels	O
.	O
</s>
<s>
The	O
"	O
V	O
"	O
model	O
of	O
big	B-Application
data	I-Application
is	O
concerning	O
as	O
it	O
centers	O
around	O
computational	O
scalability	O
and	O
lacks	O
in	O
a	O
loss	O
around	O
the	O
perceptibility	O
and	O
understandability	O
of	O
information	O
.	O
</s>
<s>
This	O
led	O
to	O
the	O
framework	O
of	O
cognitive	O
big	B-Application
data	I-Application
,	O
which	O
characterizes	O
big	B-Application
data	I-Application
applications	O
according	O
to	O
:	O
</s>
<s>
Large	O
data	B-General_Concept
sets	I-General_Concept
have	O
been	O
analyzed	O
by	O
computing	O
machines	O
for	O
well	O
over	O
a	O
century	O
,	O
including	O
the	O
US	O
census	O
analytics	O
performed	O
by	O
IBM	O
's	O
punch-card	O
machines	O
which	O
computed	O
statistics	O
including	O
means	O
and	O
variances	O
of	O
populations	O
across	O
the	O
whole	O
continent	O
.	O
</s>
<s>
In	O
more	O
recent	O
decades	O
,	O
science	O
experiments	O
such	O
as	O
CERN	O
have	O
produced	O
data	O
on	O
similar	O
scales	O
to	O
current	O
commercial	O
"	O
big	B-Application
data	I-Application
"	O
.	O
</s>
<s>
However	O
,	O
science	O
experiments	O
have	O
tended	O
to	O
analyze	O
their	O
data	O
using	O
specialized	O
custom-built	O
high-performance	B-Architecture
computing	I-Architecture
(	O
super-computing	B-Architecture
)	O
clusters	O
and	O
grids	O
,	O
rather	O
than	O
clouds	O
of	O
cheap	O
commodity	O
computers	O
as	O
in	O
the	O
current	O
commercial	O
wave	O
,	O
implying	O
a	O
difference	O
in	O
both	O
culture	O
and	O
technology	O
stack	O
.	O
</s>
<s>
Ulf-Dietrich	O
Reips	O
and	O
Uwe	O
Matzat	O
wrote	O
in	O
2014	O
that	O
big	B-Application
data	I-Application
had	O
become	O
a	O
"	O
fad	O
"	O
in	O
scientific	O
research	O
.	O
</s>
<s>
Researcher	O
danah	O
boyd	O
has	O
raised	O
concerns	O
about	O
the	O
use	O
of	O
big	B-Application
data	I-Application
in	O
science	O
neglecting	O
principles	O
such	O
as	O
choosing	O
a	O
representative	O
sample	O
by	O
being	O
too	O
concerned	O
about	O
handling	O
the	O
huge	O
amounts	O
of	O
data	O
.	O
</s>
<s>
Integration	B-General_Concept
across	O
heterogeneous	O
data	O
resources	O
—	O
some	O
that	O
might	O
be	O
considered	O
big	B-Application
data	I-Application
and	O
others	O
not	O
—	O
presents	O
formidable	O
logistical	O
as	O
well	O
as	O
analytical	O
challenges	O
,	O
but	O
many	O
researchers	O
argue	O
that	O
such	O
integrations	O
are	O
likely	O
to	O
represent	O
the	O
most	O
promising	O
new	O
frontiers	O
in	O
science	O
.	O
</s>
<s>
In	O
the	O
provocative	O
article	O
"	O
Critical	O
Questions	O
for	O
Big	B-Application
Data	I-Application
"	O
,	O
the	O
authors	O
title	O
big	B-Application
data	I-Application
a	O
part	O
of	O
mythology	O
:	O
"	O
large	O
data	B-General_Concept
sets	I-General_Concept
offer	O
a	O
higher	O
form	O
of	O
intelligence	O
and	O
knowledge	O
[...],	O
with	O
the	O
aura	O
of	O
truth	O
,	O
objectivity	O
,	O
and	O
accuracy	O
"	O
.	O
</s>
<s>
Users	O
of	O
big	B-Application
data	I-Application
are	O
often	O
"	O
lost	O
in	O
the	O
sheer	O
volume	O
of	O
numbers	O
"	O
,	O
and	O
"	O
working	O
with	O
Big	B-Application
Data	I-Application
is	O
still	O
subjective	O
,	O
and	O
what	O
it	O
quantifies	O
does	O
not	O
necessarily	O
have	O
a	O
closer	O
claim	O
on	O
objective	O
truth	O
"	O
.	O
</s>
<s>
Recent	O
developments	O
in	O
BI	O
domain	O
,	O
such	O
as	O
pro-active	O
reporting	O
especially	O
target	O
improvements	O
in	O
the	O
usability	O
of	O
big	B-Application
data	I-Application
,	O
through	O
automated	O
filtering	B-Application
of	O
non-useful	O
data	O
and	O
correlations	O
.	O
</s>
<s>
Catherine	O
Tucker	O
has	O
pointed	O
to	O
"	O
hype	O
"	O
around	O
big	B-Application
data	I-Application
,	O
writing	O
"	O
By	O
itself	O
,	O
big	B-Application
data	I-Application
is	O
unlikely	O
to	O
be	O
valuable.	O
"	O
</s>
<s>
Big	B-Application
data	I-Application
analysis	I-Application
is	O
often	O
shallow	O
compared	O
to	O
analysis	O
of	O
smaller	O
data	B-General_Concept
sets	I-General_Concept
.	O
</s>
<s>
In	O
many	O
big	B-Application
data	I-Application
projects	O
,	O
there	O
is	O
no	O
large	O
data	B-General_Concept
analysis	I-General_Concept
happening	O
,	O
but	O
the	O
challenge	O
is	O
the	O
extract	O
,	O
transform	O
,	O
load	O
part	O
of	O
data	O
pre-processing	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
is	O
a	O
buzzword	O
and	O
a	O
"	O
vague	O
term	O
"	O
,	O
but	O
at	O
the	O
same	O
time	O
an	O
"	O
obsession	O
"	O
with	O
entrepreneurs	O
,	O
consultants	O
,	O
scientists	O
,	O
and	O
the	O
media	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
showcases	O
such	O
as	O
Google	B-Application
Flu	O
Trends	O
failed	O
to	O
deliver	O
good	O
predictions	O
in	O
recent	O
years	O
,	O
overstating	O
the	O
flu	O
outbreaks	O
by	O
a	O
factor	O
of	O
two	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
often	O
poses	O
the	O
same	O
challenges	O
as	O
small	B-General_Concept
data	I-General_Concept
;	O
adding	O
more	O
data	O
does	O
not	O
solve	O
problems	O
of	O
bias	O
,	O
but	O
may	O
emphasize	O
other	O
problems	O
.	O
</s>
<s>
Google	B-Application
Translate	I-Application
—	O
which	O
is	O
based	O
on	O
big	B-Application
data	I-Application
statistical	O
analysis	O
of	O
text	O
—	O
does	O
a	O
good	O
job	O
at	O
translating	O
web	O
pages	O
.	O
</s>
<s>
On	O
the	O
other	O
hand	O
,	O
big	B-Application
data	I-Application
may	O
also	O
introduce	O
new	O
problems	O
,	O
such	O
as	O
the	O
multiple	B-General_Concept
comparisons	I-General_Concept
problem	I-General_Concept
:	O
simultaneously	O
testing	O
a	O
large	O
set	O
of	O
hypotheses	O
is	O
likely	O
to	O
produce	O
many	O
false	O
results	O
that	O
mistakenly	O
appear	O
significant	O
.	O
</s>
<s>
process	O
a	O
big	O
amount	O
of	O
scientific	O
data	O
;	O
although	O
not	O
with	O
big	B-Application
data	I-Application
technology	O
)	O
,	O
the	O
likelihood	O
of	O
a	O
"	O
significant	O
"	O
result	O
being	O
false	O
grows	O
fast	O
–	O
even	O
more	O
so	O
,	O
when	O
only	O
positive	O
results	O
are	O
published	O
.	O
</s>
<s>
Furthermore	O
,	O
big	B-Application
data	I-Application
analytics	I-Application
results	O
are	O
only	O
as	O
good	O
as	O
the	O
model	O
on	O
which	O
they	O
are	O
predicated	O
.	O
</s>
<s>
In	O
an	O
example	O
,	O
big	B-Application
data	I-Application
took	O
part	O
in	O
attempting	O
to	O
predict	O
the	O
results	O
of	O
the	O
2016	O
U.S.	O
Presidential	O
Election	O
with	O
varying	O
degrees	O
of	O
success	O
.	O
</s>
<s>
Big	B-Application
data	I-Application
has	O
been	O
used	O
in	O
policing	O
and	O
surveillance	O
by	O
institutions	O
like	O
law	O
enforcement	O
and	O
corporations	O
.	O
</s>
<s>
Due	O
to	O
the	O
less	O
visible	O
nature	O
of	O
data-based	O
surveillance	O
as	O
compared	O
to	O
traditional	O
methods	O
of	O
policing	O
,	O
objections	O
to	O
big	B-Application
data	I-Application
policing	O
are	O
less	O
likely	O
to	O
arise	O
.	O
</s>
<s>
According	O
to	O
Sarah	O
Brayne	O
's	O
Big	B-Application
Data	I-Application
Surveillance	O
:	O
The	O
Case	O
of	O
Policing	O
,	O
big	B-Application
data	I-Application
policing	O
can	O
reproduce	O
existing	O
societal	O
inequalities	O
in	O
three	O
ways	O
:	O
</s>
<s>
If	O
these	O
potential	O
problems	O
are	O
not	O
corrected	O
or	O
regulated	O
,	O
the	O
effects	O
of	O
big	B-Application
data	I-Application
policing	O
may	O
continue	O
to	O
shape	O
societal	O
hierarchies	O
.	O
</s>
<s>
Conscientious	O
usage	O
of	O
big	B-Application
data	I-Application
policing	O
could	O
prevent	O
individual	O
level	O
biases	O
from	O
becoming	O
institutional	O
biases	O
,	O
Brayne	O
also	O
notes	O
.	O
</s>
