<s>
A	O
Web	B-Application
crawler	I-Application
,	O
sometimes	O
called	O
a	O
spider	O
or	O
spiderbot	B-Application
and	O
often	O
shortened	O
to	O
crawler	O
,	O
is	O
an	O
Internet	B-Protocol
bot	I-Protocol
that	O
systematically	O
browses	O
the	O
World	O
Wide	O
Web	O
and	O
that	O
is	O
typically	O
operated	O
by	O
search	B-Application
engines	I-Application
for	O
the	O
purpose	O
of	O
Web	O
indexing	O
(	O
web	O
spidering	B-Application
)	O
.	O
</s>
<s>
Web	B-Application
search	I-Application
engines	I-Application
and	O
some	O
other	O
websites	O
use	O
Web	B-Application
crawling	I-Application
or	O
spidering	B-Application
software	O
to	O
update	O
their	O
web	O
content	O
or	O
indices	O
of	O
other	O
sites	O
 '	O
web	O
content	O
.	O
</s>
<s>
Web	B-Application
crawlers	I-Application
copy	O
pages	O
for	O
processing	O
by	O
a	O
search	B-Application
engine	I-Application
,	O
which	O
indexes	B-Application
the	O
downloaded	O
pages	O
so	O
that	O
users	O
can	O
search	O
more	O
efficiently	O
.	O
</s>
<s>
For	O
example	O
,	O
including	O
a	O
robots.txt	O
file	O
can	O
request	O
bots	B-General_Concept
to	O
index	O
only	O
parts	O
of	O
a	O
website	O
,	O
or	O
nothing	O
at	O
all	O
.	O
</s>
<s>
For	O
this	O
reason	O
,	O
search	B-Application
engines	I-Application
struggled	O
to	O
give	O
relevant	O
search	B-Application
results	I-Application
in	O
the	O
early	O
years	O
of	O
the	O
World	O
Wide	O
Web	O
,	O
before	O
2000	O
.	O
</s>
<s>
Crawlers	O
can	O
validate	O
hyperlinks	O
and	O
HTML	B-Language
code	I-Language
.	O
</s>
<s>
They	O
can	O
also	O
be	O
used	O
for	O
web	B-Application
scraping	I-Application
and	O
data-driven	B-Application
programming	I-Application
.	O
</s>
<s>
A	O
web	B-Application
crawler	I-Application
is	O
also	O
known	O
as	O
a	O
spider	O
,	O
an	O
ant	O
,	O
an	O
automatic	B-Application
indexer	I-Application
,	O
or	O
(	O
in	O
the	O
FOAF	O
software	O
context	O
)	O
a	O
Web	B-Application
scutter	I-Application
.	O
</s>
<s>
A	O
Web	B-Application
crawler	I-Application
starts	O
with	O
a	O
list	O
of	O
URLs	O
to	O
visit	O
.	O
</s>
<s>
As	O
the	O
crawler	O
visits	O
these	O
URLs	O
,	O
by	O
communicating	O
with	O
web	B-Application
servers	I-Application
that	O
respond	O
to	O
those	O
URLs	O
,	O
it	O
identifies	O
all	O
the	O
hyperlinks	O
in	O
the	O
retrieved	O
web	O
pages	O
and	O
adds	O
them	O
to	O
the	O
list	O
of	O
URLs	O
to	O
visit	O
,	O
called	O
the	O
crawl	B-Application
frontier	I-Application
.	O
</s>
<s>
The	O
archive	O
is	O
known	O
as	O
the	O
repository	B-General_Concept
and	O
is	O
designed	O
to	O
store	O
and	O
manage	O
the	O
collection	O
of	O
web	O
pages	O
.	O
</s>
<s>
The	O
repository	B-General_Concept
only	O
stores	O
HTML	B-Language
pages	O
and	O
these	O
pages	O
are	O
stored	O
as	O
distinct	O
files	O
.	O
</s>
<s>
A	O
repository	B-General_Concept
is	O
similar	O
to	O
any	O
other	O
system	O
that	O
stores	O
data	O
,	O
like	O
a	O
modern-day	O
database	O
.	O
</s>
<s>
The	O
only	O
difference	O
is	O
that	O
a	O
repository	B-General_Concept
does	O
not	O
need	O
all	O
the	O
functionality	O
offered	O
by	O
a	O
database	O
system	O
.	O
</s>
<s>
The	O
repository	B-General_Concept
stores	O
the	O
most	O
recent	O
version	O
of	O
the	O
web	O
page	O
retrieved	O
by	O
the	O
crawler	O
.	O
</s>
<s>
The	O
number	O
of	O
possible	O
URLs	O
crawled	O
being	O
generated	O
by	O
server-side	O
software	O
has	O
also	O
made	O
it	O
difficult	O
for	O
web	B-Application
crawlers	I-Application
to	O
avoid	O
retrieving	O
duplicate	O
content	O
.	O
</s>
<s>
Endless	O
combinations	O
of	O
HTTP	B-Protocol
GET	O
(	O
URL-based	O
)	O
parameters	O
exist	O
,	O
of	O
which	O
only	O
a	O
small	O
selection	O
will	O
actually	O
return	O
unique	O
content	O
.	O
</s>
<s>
For	O
example	O
,	O
a	O
simple	O
online	O
photo	O
gallery	O
may	O
offer	O
three	O
options	O
to	O
users	O
,	O
as	O
specified	O
through	O
HTTP	B-Protocol
GET	O
parameters	O
in	O
the	O
URL	O
.	O
</s>
<s>
The	O
behavior	O
of	O
a	O
Web	B-Application
crawler	I-Application
is	O
the	O
outcome	O
of	O
a	O
combination	O
of	O
policies	O
:	O
</s>
<s>
a	O
parallelization	B-Operating_System
policy	O
that	O
states	O
how	O
to	O
coordinate	O
distributed	O
web	B-Application
crawlers	I-Application
.	O
</s>
<s>
Given	O
the	O
current	O
size	O
of	O
the	O
Web	O
,	O
even	O
large	O
search	B-Application
engines	I-Application
cover	O
only	O
a	O
portion	O
of	O
the	O
publicly	O
available	O
part	O
.	O
</s>
<s>
A	O
2009	O
study	O
showed	O
even	O
large-scale	O
search	B-Application
engines	I-Application
index	O
no	O
more	O
than	O
40	O
–	O
70%	O
of	O
the	O
indexable	O
Web	O
;	O
a	O
previous	O
study	O
by	O
Steve	O
Lawrence	O
and	O
Lee	O
Giles	O
showed	O
that	O
no	O
search	B-Application
engine	I-Application
indexed	I-Application
more	O
than	O
16%	O
of	O
the	O
Web	O
in	O
1999	O
.	O
</s>
<s>
The	O
importance	O
of	O
a	O
page	O
is	O
a	O
function	O
of	O
its	O
intrinsic	O
quality	O
,	O
its	O
popularity	O
in	O
terms	O
of	O
links	O
or	O
visits	O
,	O
and	O
even	O
of	O
its	O
URL	O
(	O
the	O
latter	O
is	O
the	O
case	O
of	O
vertical	B-Application
search	I-Application
engines	I-Application
restricted	O
to	O
a	O
single	O
top-level	O
domain	O
,	O
or	O
search	B-Application
engines	I-Application
restricted	O
to	O
a	O
fixed	O
Web	O
site	O
)	O
.	O
</s>
<s>
The	O
ordering	O
metrics	O
tested	O
were	O
breadth-first	B-Algorithm
,	O
backlink	O
count	O
and	O
partial	O
PageRank	B-Algorithm
calculations	O
.	O
</s>
<s>
One	O
of	O
the	O
conclusions	O
was	O
that	O
if	O
the	O
crawler	O
wants	O
to	O
download	O
pages	O
with	O
high	O
Pagerank	B-Algorithm
early	O
during	O
the	O
crawling	O
process	O
,	O
then	O
the	O
partial	O
Pagerank	B-Algorithm
strategy	O
is	O
the	O
better	O
,	O
followed	O
by	O
breadth-first	B-Algorithm
and	O
backlink-count	O
.	O
</s>
<s>
Cho	O
also	O
wrote	O
his	O
PhD	O
dissertation	O
at	O
Stanford	O
on	O
web	B-Application
crawling	I-Application
.	O
</s>
<s>
Najork	O
and	O
Wiener	O
performed	O
an	O
actual	O
crawl	O
on	O
328	O
million	O
pages	O
,	O
using	O
breadth-first	B-Algorithm
ordering	O
.	O
</s>
<s>
They	O
found	O
that	O
a	O
breadth-first	B-Algorithm
crawl	O
captures	O
pages	O
with	O
high	O
Pagerank	B-Algorithm
early	O
in	O
the	O
crawl	O
(	O
but	O
they	O
did	O
not	O
compare	O
this	O
strategy	O
against	O
other	O
strategies	O
)	O
.	O
</s>
<s>
It	O
is	O
similar	O
to	O
a	O
PageRank	B-Algorithm
computation	O
,	O
but	O
it	O
is	O
faster	O
and	O
is	O
only	O
done	O
in	O
one	O
step	O
.	O
</s>
<s>
used	O
simulation	O
on	O
subsets	O
of	O
the	O
Web	O
of	O
40	O
million	O
pages	O
from	O
the	O
.it	O
domain	O
and	O
100	O
million	O
pages	O
from	O
the	O
WebBase	O
crawl	O
,	O
testing	O
breadth-first	B-Algorithm
against	O
depth-first	O
,	O
random	O
ordering	O
and	O
an	O
omniscient	O
strategy	O
.	O
</s>
<s>
The	O
comparison	O
was	O
based	O
on	O
how	O
well	O
PageRank	B-Algorithm
computed	O
on	O
a	O
partial	O
crawl	O
approximates	O
the	O
true	O
PageRank	B-Algorithm
value	O
.	O
</s>
<s>
Surprisingly	O
,	O
some	O
visits	O
that	O
accumulate	O
PageRank	B-Algorithm
very	O
quickly	O
(	O
most	O
notably	O
,	O
breadth-first	B-Algorithm
and	O
the	O
omniscient	O
visit	O
)	O
provide	O
very	O
poor	O
progressive	O
approximations	O
.	O
</s>
<s>
They	O
showed	O
that	O
both	O
the	O
OPIC	O
strategy	O
and	O
a	O
strategy	O
that	O
uses	O
the	O
length	O
of	O
the	O
per-site	O
queues	O
are	O
better	O
than	O
breadth-first	B-Algorithm
crawling	O
,	O
and	O
that	O
it	O
is	O
also	O
very	O
effective	O
to	O
use	O
a	O
previous	O
crawl	O
,	O
when	O
it	O
is	O
available	O
,	O
to	O
guide	O
the	O
current	O
one	O
.	O
</s>
<s>
Their	O
method	O
crawls	O
web	O
pages	O
with	O
high	O
PageRank	B-Algorithm
from	O
different	O
communities	O
in	O
less	O
iteration	O
in	O
comparison	O
with	O
crawl	O
starting	O
from	O
random	O
seeds	O
.	O
</s>
<s>
A	O
crawler	O
may	O
only	O
want	O
to	O
seek	O
out	O
HTML	B-Language
pages	O
and	O
avoid	O
all	O
other	O
MIME	B-Application
types	I-Application
.	O
</s>
<s>
In	O
order	O
to	O
request	O
only	O
HTML	B-Language
resources	O
,	O
a	O
crawler	O
may	O
make	O
an	O
HTTP	B-Protocol
HEAD	O
request	O
to	O
determine	O
a	O
Web	O
resource	O
's	O
MIME	B-Application
type	I-Application
before	O
requesting	O
the	O
entire	O
resource	O
with	O
a	O
GET	O
request	O
.	O
</s>
<s>
To	O
avoid	O
making	O
numerous	O
HEAD	O
requests	O
,	O
a	O
crawler	O
may	O
examine	O
the	O
URL	O
and	O
only	O
request	O
a	O
resource	O
if	O
the	O
URL	O
ends	O
with	O
certain	O
characters	O
such	O
as	O
.html	B-Language
,	O
.htm	B-Language
,	O
.asp	O
,	O
.aspx	O
,	O
.php	O
,	O
.jsp	O
,	O
.jspx	O
or	O
a	O
slash	O
.	O
</s>
<s>
This	O
strategy	O
may	O
cause	O
numerous	O
HTML	B-Language
Web	O
resources	O
to	O
be	O
unintentionally	O
skipped	O
.	O
</s>
<s>
Some	O
crawlers	O
may	O
also	O
avoid	O
requesting	O
any	O
resources	O
that	O
have	O
a	O
"	B-Protocol
?	I-Protocol
"	I-Protocol
</s>
<s>
This	O
strategy	O
is	O
unreliable	O
if	O
the	O
site	O
uses	O
URL	B-Protocol
rewriting	I-Protocol
to	O
simplify	O
its	O
URLs	O
.	O
</s>
<s>
Crawlers	O
usually	O
perform	O
some	O
type	O
of	O
URL	B-Protocol
normalization	I-Protocol
in	O
order	O
to	O
avoid	O
crawling	O
the	O
same	O
resource	O
more	O
than	O
once	O
.	O
</s>
<s>
The	O
term	O
URL	B-Protocol
normalization	I-Protocol
,	O
also	O
called	O
URL	B-Protocol
canonicalization	I-Protocol
,	O
refers	O
to	O
the	O
process	O
of	O
modifying	O
and	O
standardizing	O
a	O
URL	O
in	O
a	O
consistent	O
manner	O
.	O
</s>
<s>
For	O
example	O
,	O
when	O
given	O
a	O
seed	O
URL	O
of	O
http://llama.org/hamster/monkey/page.html	O
,	O
it	O
will	O
attempt	O
to	O
crawl	O
/hamster/monkey/	O
,	O
/hamster/	O
,	O
and	O
/	O
.	O
</s>
<s>
Web	B-Application
crawlers	I-Application
that	O
attempt	O
to	O
download	O
pages	O
that	O
are	O
similar	O
to	O
each	O
other	O
are	O
called	O
focused	B-Protocol
crawler	I-Protocol
or	O
topical	B-Protocol
crawlers	I-Protocol
.	O
</s>
<s>
The	O
concepts	O
of	O
topical	O
and	O
focused	B-Protocol
crawling	I-Protocol
were	O
first	O
introduced	O
by	O
Filippo	O
Menczer	O
and	O
by	O
Soumen	O
Chakrabarti	O
et	O
al	O
.	O
</s>
<s>
The	O
main	O
problem	O
in	O
focused	B-Protocol
crawling	I-Protocol
is	O
that	O
in	O
the	O
context	O
of	O
a	O
Web	B-Application
crawler	I-Application
,	O
we	O
would	O
like	O
to	O
be	O
able	O
to	O
predict	O
the	O
similarity	O
of	O
the	O
text	O
of	O
a	O
given	O
page	O
to	O
the	O
query	O
before	O
actually	O
downloading	O
the	O
page	O
.	O
</s>
<s>
A	O
possible	O
predictor	O
is	O
the	O
anchor	O
text	O
of	O
links	O
;	O
this	O
was	O
the	O
approach	O
taken	O
by	O
Pinkerton	O
in	O
the	O
first	O
web	B-Application
crawler	I-Application
of	O
the	O
early	O
days	O
of	O
the	O
Web	O
.	O
</s>
<s>
The	O
performance	O
of	O
a	O
focused	B-Protocol
crawling	I-Protocol
depends	O
mostly	O
on	O
the	O
richness	O
of	O
links	O
in	O
the	O
specific	O
topic	O
being	O
searched	O
,	O
and	O
a	O
focused	B-Protocol
crawling	I-Protocol
usually	O
relies	O
on	O
a	O
general	O
Web	B-Application
search	I-Application
engine	I-Application
for	O
providing	O
starting	O
points	O
.	O
</s>
<s>
An	O
example	O
of	O
the	O
focused	B-Protocol
crawlers	I-Protocol
are	O
academic	O
crawlers	O
,	O
which	O
crawls	O
free-access	O
academic	O
related	O
documents	O
,	O
such	O
as	O
the	O
citeseerxbot	O
,	O
which	O
is	O
the	O
crawler	O
of	O
CiteSeerX	B-Application
search	B-Application
engine	I-Application
.	O
</s>
<s>
Other	O
academic	O
search	B-Application
engines	I-Application
are	O
Google	B-Library
Scholar	I-Library
and	O
Microsoft	B-Application
Academic	I-Application
Search	I-Application
etc	O
.	O
</s>
<s>
Because	O
most	O
academic	O
papers	O
are	O
published	O
in	O
PDF	B-Application
formats	I-Application
,	O
such	O
kind	O
of	O
crawler	O
is	O
particularly	O
interested	O
in	O
crawling	O
PDF	B-Application
,	O
PostScript	B-Language
files	O
,	O
Microsoft	B-Application
Word	I-Application
including	O
their	O
zipped	B-General_Concept
formats	O
.	O
</s>
<s>
Because	O
of	O
this	O
,	O
general	O
open-source	O
crawlers	O
,	O
such	O
as	O
Heritrix	B-Application
,	O
must	O
be	O
customized	O
to	O
filter	O
out	O
other	O
MIME	B-Application
types	I-Application
,	O
or	O
a	O
middleware	B-General_Concept
is	O
used	O
to	O
extract	O
these	O
documents	O
out	O
and	O
import	O
them	O
to	O
the	O
focused	O
crawl	O
database	O
and	O
repository	B-General_Concept
.	O
</s>
<s>
Identifying	O
whether	O
these	O
documents	O
are	O
academic	O
or	O
not	O
is	O
challenging	O
and	O
can	O
add	O
a	O
significant	O
overhead	O
to	O
the	O
crawling	O
process	O
,	O
so	O
this	O
is	O
performed	O
as	O
a	O
post	O
crawling	O
process	O
using	O
machine	O
learning	O
or	O
regular	B-Language
expression	I-Language
algorithms	O
.	O
</s>
<s>
Because	O
academic	O
documents	O
make	O
up	O
only	O
a	O
small	O
fraction	O
of	O
all	O
web	O
pages	O
,	O
a	O
good	O
seed	O
selection	O
is	O
important	O
in	O
boosting	O
the	O
efficiencies	O
of	O
these	O
web	B-Application
crawlers	I-Application
.	O
</s>
<s>
Other	O
academic	O
crawlers	O
may	O
download	O
plain	O
text	O
and	O
HTML	B-Language
files	I-Language
,	O
that	O
contains	O
metadata	O
of	O
academic	O
papers	O
,	O
such	O
as	O
titles	O
,	O
papers	O
,	O
and	O
abstracts	O
.	O
</s>
<s>
This	O
increases	O
the	O
overall	O
number	O
of	O
papers	O
,	O
but	O
a	O
significant	O
fraction	O
may	O
not	O
provide	O
free	O
PDF	B-Application
downloads	O
.	O
</s>
<s>
Another	O
type	O
of	O
focused	B-Protocol
crawlers	I-Protocol
is	O
semantic	O
focused	B-Protocol
crawler	I-Protocol
,	O
which	O
makes	O
use	O
of	O
domain	O
ontologies	O
to	O
represent	O
topical	O
maps	O
and	O
link	O
Web	O
pages	O
with	O
relevant	O
ontological	O
concepts	O
for	O
the	O
selection	O
and	O
categorization	O
purposes	O
.	O
</s>
<s>
introduced	O
such	O
an	O
ontology-learning-based	O
crawler	O
using	O
a	O
support-vector	B-Algorithm
machine	I-Algorithm
to	O
update	O
the	O
content	O
of	O
ontological	O
concepts	O
when	O
crawling	O
Web	O
pages	O
.	O
</s>
<s>
By	O
the	O
time	O
a	O
Web	B-Application
crawler	I-Application
has	O
finished	O
its	O
crawl	O
,	O
many	O
events	O
could	O
have	O
happened	O
,	O
including	O
creations	O
,	O
updates	O
,	O
and	O
deletions	O
.	O
</s>
<s>
From	O
the	O
search	B-Application
engine	I-Application
's	O
point	O
of	O
view	O
,	O
there	O
is	O
a	O
cost	O
associated	O
with	O
not	O
detecting	O
an	O
event	O
,	O
and	O
thus	O
having	O
an	O
outdated	O
copy	O
of	O
a	O
resource	O
.	O
</s>
<s>
The	O
freshness	O
of	O
a	O
page	O
p	O
in	O
the	O
repository	B-General_Concept
at	O
time	O
t	O
is	O
defined	O
as	O
:	O
</s>
<s>
The	O
age	O
of	O
a	O
page	O
p	O
in	O
the	O
repository	B-General_Concept
,	O
at	O
time	O
t	O
is	O
defined	O
as	O
:	O
</s>
<s>
worked	O
with	O
a	O
definition	O
of	O
the	O
objective	O
of	O
a	O
Web	B-Application
crawler	I-Application
that	O
is	O
equivalent	O
to	O
freshness	O
,	O
but	O
use	O
a	O
different	O
wording	O
:	O
they	O
propose	O
that	O
a	O
crawler	O
must	O
minimize	O
the	O
fraction	O
of	O
time	O
pages	O
remain	O
outdated	O
.	O
</s>
<s>
They	O
also	O
noted	O
that	O
the	O
problem	O
of	O
Web	B-Application
crawling	I-Application
can	O
be	O
modeled	O
as	O
a	O
multiple-queue	O
,	O
single-server	O
polling	O
system	O
,	O
on	O
which	O
the	O
Web	B-Application
crawler	I-Application
is	O
the	O
server	O
and	O
the	O
Web	O
sites	O
are	O
the	O
queues	O
.	O
</s>
<s>
Under	O
this	O
model	O
,	O
mean	O
waiting	O
time	O
for	O
a	O
customer	O
in	O
the	O
polling	O
system	O
is	O
equivalent	O
to	O
the	O
average	O
age	O
for	O
the	O
Web	B-Application
crawler	I-Application
.	O
</s>
<s>
Intuitively	O
,	O
the	O
reasoning	O
is	O
that	O
,	O
as	O
web	B-Application
crawlers	I-Application
have	O
a	O
limit	O
to	O
how	O
many	O
pages	O
they	O
can	O
crawl	O
in	O
a	O
given	O
time	O
frame	O
,	O
(	O
1	O
)	O
they	O
will	O
allocate	O
too	O
many	O
new	O
crawls	O
to	O
rapidly	O
changing	O
pages	O
at	O
the	O
expense	O
of	O
less	O
frequently	O
updating	O
pages	O
,	O
and	O
(	O
2	O
)	O
the	O
freshness	O
of	O
rapidly	O
changing	O
pages	O
lasts	O
for	O
shorter	O
period	O
than	O
that	O
of	O
less	O
frequently	O
changing	O
pages	O
.	O
</s>
<s>
As	O
noted	O
by	O
Koster	O
,	O
the	O
use	O
of	O
Web	B-Application
crawlers	I-Application
is	O
useful	O
for	O
a	O
number	O
of	O
tasks	O
,	O
but	O
comes	O
with	O
a	O
price	O
for	O
the	O
general	O
community	O
.	O
</s>
<s>
The	O
costs	O
of	O
using	O
Web	B-Application
crawlers	I-Application
include	O
:	O
</s>
<s>
network	O
resources	O
,	O
as	O
crawlers	O
require	O
considerable	O
bandwidth	O
and	O
operate	O
with	O
a	O
high	O
degree	O
of	O
parallelism	B-Operating_System
during	O
a	O
long	O
period	O
of	O
time	O
;	O
</s>
<s>
personal	O
crawlers	O
that	O
,	O
if	O
deployed	O
by	O
too	O
many	O
users	O
,	O
can	O
disrupt	O
networks	O
and	O
Web	B-Application
servers	I-Application
.	O
</s>
<s>
A	O
partial	O
solution	O
to	O
these	O
problems	O
is	O
the	O
robots	O
exclusion	O
protocol	O
,	O
also	O
known	O
as	O
the	O
robots.txt	O
protocol	O
that	O
is	O
a	O
standard	O
for	O
administrators	O
to	O
indicate	O
which	O
parts	O
of	O
their	O
Web	B-Application
servers	I-Application
should	O
not	O
be	O
accessed	O
by	O
crawlers	O
.	O
</s>
<s>
Recently	O
commercial	O
search	B-Application
engines	I-Application
like	O
Google	B-Application
,	O
Ask	B-Application
Jeeves	I-Application
,	O
MSN	B-Application
and	O
Yahoo	B-Application
!	I-Application
</s>
<s>
However	O
,	O
if	O
pages	O
were	O
downloaded	O
at	O
this	O
rate	O
from	O
a	O
website	O
with	O
more	O
than	O
100,000	O
pages	O
over	O
a	O
perfect	O
connection	O
with	O
zero	O
latency	O
and	O
infinite	O
bandwidth	O
,	O
it	O
would	O
take	O
more	O
than	O
2	O
months	O
to	O
download	O
only	O
that	O
entire	O
Web	O
site	O
;	O
also	O
,	O
only	O
a	O
fraction	O
of	O
the	O
resources	O
from	O
that	O
Web	B-Application
server	I-Application
would	O
be	O
used	O
.	O
</s>
<s>
For	O
those	O
using	O
Web	B-Application
crawlers	I-Application
for	O
research	O
purposes	O
,	O
a	O
more	O
detailed	O
cost-benefit	O
analysis	O
is	O
needed	O
and	O
ethical	O
considerations	O
should	O
be	O
taken	O
into	O
account	O
when	O
deciding	O
where	O
to	O
crawl	O
and	O
how	O
fast	O
to	O
crawl	O
.	O
</s>
<s>
It	O
is	O
worth	O
noticing	O
that	O
even	O
when	O
being	O
very	O
polite	O
,	O
and	O
taking	O
all	O
the	O
safeguards	O
to	O
avoid	O
overloading	O
Web	B-Application
servers	I-Application
,	O
some	O
complaints	O
from	O
Web	B-Application
server	I-Application
administrators	O
are	O
received	O
.	O
</s>
<s>
A	O
parallel	B-Operating_System
crawler	O
is	O
a	O
crawler	O
that	O
runs	O
multiple	O
processes	O
in	O
parallel	B-Operating_System
.	O
</s>
<s>
The	O
goal	O
is	O
to	O
maximize	O
the	O
download	O
rate	O
while	O
minimizing	O
the	O
overhead	O
from	O
parallelization	B-Operating_System
and	O
to	O
avoid	O
repeated	O
downloads	O
of	O
the	O
same	O
page	O
.	O
</s>
<s>
Web	B-Application
crawlers	I-Application
are	O
a	O
central	O
part	O
of	O
search	B-Application
engines	I-Application
,	O
and	O
details	O
on	O
their	O
algorithms	O
and	O
architecture	O
are	O
kept	O
as	O
business	O
secrets	O
.	O
</s>
<s>
There	O
are	O
also	O
emerging	O
concerns	O
about	O
"	O
search	B-Application
engine	I-Application
spamming	O
"	O
,	O
which	O
prevent	O
major	O
search	B-Application
engines	I-Application
from	O
publishing	O
their	O
ranking	O
algorithms	O
.	O
</s>
<s>
While	O
most	O
of	O
the	O
website	O
owners	O
are	O
keen	O
to	O
have	O
their	O
pages	O
indexed	O
as	O
broadly	O
as	O
possible	O
to	O
have	O
strong	O
presence	O
in	O
search	B-Application
engines	I-Application
,	O
web	B-Application
crawling	I-Application
can	O
also	O
have	O
unintended	O
consequences	O
and	O
lead	O
to	O
a	O
compromise	O
or	O
data	O
breach	O
if	O
a	O
search	B-Application
engine	I-Application
indexes	B-Application
resources	O
that	O
should	O
n't	O
be	O
publicly	O
available	O
,	O
or	O
pages	O
revealing	O
potentially	O
vulnerable	O
versions	O
of	O
software	O
.	O
</s>
<s>
Apart	O
from	O
standard	O
web	O
application	O
security	O
recommendations	O
website	O
owners	O
can	O
reduce	O
their	O
exposure	O
to	O
opportunistic	O
hacking	O
by	O
only	O
allowing	O
search	B-Application
engines	I-Application
to	O
index	O
the	O
public	O
parts	O
of	O
their	O
websites	O
(	O
with	O
robots.txt	O
)	O
and	O
explicitly	O
blocking	O
them	O
from	O
indexing	O
transactional	O
parts	O
(	O
login	O
pages	O
,	O
private	O
pages	O
,	O
etc	O
.	O
</s>
<s>
Web	B-Application
crawlers	I-Application
typically	O
identify	O
themselves	O
to	O
a	O
Web	B-Application
server	I-Application
by	O
using	O
the	O
User-agent	B-Protocol
field	O
of	O
an	O
HTTP	B-Protocol
request	O
.	O
</s>
<s>
Web	O
site	O
administrators	O
typically	O
examine	O
their	O
Web	B-Application
servers	I-Application
 '	O
log	O
and	O
use	O
the	O
user	B-Protocol
agent	I-Protocol
field	O
to	O
determine	O
which	O
crawlers	O
have	O
visited	O
the	O
web	B-Application
server	I-Application
and	O
how	O
often	O
.	O
</s>
<s>
The	O
user	B-Protocol
agent	I-Protocol
field	O
may	O
include	O
a	O
URL	O
where	O
the	O
Web	O
site	O
administrator	O
may	O
find	O
out	O
more	O
information	O
about	O
the	O
crawler	O
.	O
</s>
<s>
Examining	O
Web	B-Application
server	I-Application
log	O
is	O
tedious	O
task	O
,	O
and	O
therefore	O
some	O
administrators	O
use	O
tools	O
to	O
identify	O
,	O
track	O
and	O
verify	O
Web	B-Application
crawlers	I-Application
.	O
</s>
<s>
Spambots	B-Protocol
and	O
other	O
malicious	O
Web	B-Application
crawlers	I-Application
are	O
unlikely	O
to	O
place	O
identifying	O
information	O
in	O
the	O
user	B-Protocol
agent	I-Protocol
field	O
,	O
or	O
they	O
may	O
mask	O
their	O
identity	O
as	O
a	O
browser	O
or	O
other	O
well-known	O
crawler	O
.	O
</s>
<s>
Web	O
site	O
administrators	O
prefer	O
Web	B-Application
crawlers	I-Application
to	O
identify	O
themselves	O
so	O
that	O
they	O
can	O
contact	O
the	O
owner	O
if	O
needed	O
.	O
</s>
<s>
In	O
some	O
cases	O
,	O
crawlers	O
may	O
be	O
accidentally	O
trapped	O
in	O
a	O
crawler	O
trap	O
or	O
they	O
may	O
be	O
overloading	O
a	O
Web	B-Application
server	I-Application
with	O
requests	O
,	O
and	O
the	O
owner	O
needs	O
to	O
stop	O
the	O
crawler	O
.	O
</s>
<s>
Identification	O
is	O
also	O
useful	O
for	O
administrators	O
that	O
are	O
interested	O
in	O
knowing	O
when	O
they	O
may	O
expect	O
their	O
Web	O
pages	O
to	O
be	O
indexed	O
by	O
a	O
particular	O
search	B-Application
engine	I-Application
.	O
</s>
<s>
Google	B-Application
's	O
Sitemaps	O
protocol	O
and	O
mod	B-Application
oai	I-Application
are	O
intended	O
to	O
allow	O
discovery	O
of	O
these	O
deep-Web	O
resources	O
.	O
</s>
<s>
Deep	O
web	B-Application
crawling	I-Application
also	O
multiplies	O
the	O
number	O
of	O
web	O
links	O
to	O
be	O
crawled	O
.	O
</s>
<s>
In	O
some	O
cases	O
,	O
such	O
as	O
the	O
Googlebot	B-Application
,	O
Web	B-Application
crawling	I-Application
is	O
done	O
on	O
all	O
text	O
contained	O
inside	O
the	O
hypertext	O
content	O
,	O
tags	O
,	O
or	O
text	O
.	O
</s>
<s>
Data	O
extracted	O
from	O
the	O
results	O
of	O
one	O
Web	O
form	O
submission	O
can	O
be	O
taken	O
and	O
applied	O
as	O
input	O
to	O
another	O
Web	O
form	O
thus	O
establishing	O
continuity	O
across	O
the	O
Deep	O
Web	O
in	O
a	O
way	O
not	O
possible	O
with	O
traditional	O
web	B-Application
crawlers	I-Application
.	O
</s>
<s>
Pages	O
built	O
on	O
AJAX	B-Device
are	O
among	O
those	O
causing	O
problems	O
to	O
web	B-Application
crawlers	I-Application
.	O
</s>
<s>
Google	B-Application
has	O
proposed	O
a	O
format	O
of	O
AJAX	B-Device
calls	O
that	O
their	O
bot	B-Protocol
can	O
recognize	O
and	O
index	O
.	O
</s>
<s>
While	O
the	O
technology	O
is	O
not	O
new	O
,	O
for	O
example	O
it	O
was	O
the	O
basis	O
of	O
Needlebase	O
which	O
has	O
been	O
bought	O
by	O
Google	B-Application
(	O
as	O
part	O
of	O
a	O
larger	O
acquisition	O
of	O
ITA	O
Labs	O
)	O
,	O
there	O
is	O
continued	O
growth	O
and	O
investment	O
in	O
this	O
area	O
by	O
investors	O
and	O
end-users	O
.	O
</s>
<s>
The	O
following	O
is	O
a	O
list	O
of	O
published	O
crawler	O
architectures	O
for	O
general-purpose	O
crawlers	O
(	O
excluding	O
focused	B-Protocol
web	I-Protocol
crawlers	I-Protocol
)	O
,	O
with	O
a	O
brief	O
description	O
that	O
includes	O
the	O
names	O
given	O
to	O
the	O
different	O
components	O
and	O
outstanding	O
features	O
:	O
</s>
<s>
World	B-Application
Wide	I-Application
Web	I-Application
Worm	I-Application
was	O
a	O
crawler	O
used	O
to	O
build	O
a	O
simple	O
index	O
of	O
document	O
titles	O
and	O
URLs	O
.	O
</s>
<s>
The	O
index	O
could	O
be	O
searched	O
by	O
using	O
the	O
grep	B-Application
Unix	B-Application
command	O
.	O
</s>
<s>
Yahoo	B-Application
!	I-Application
</s>
<s>
Slurp	O
was	O
the	O
name	O
of	O
the	O
Yahoo	B-Application
!	I-Application
</s>
<s>
Search	O
crawler	O
until	O
Yahoo	B-Application
!	I-Application
</s>
<s>
contracted	O
with	O
Microsoft	O
to	O
use	O
Bingbot	B-Protocol
instead	O
.	O
</s>
<s>
Applebot	O
is	O
Apple	O
's	O
web	B-Application
crawler	I-Application
.	O
</s>
<s>
It	O
supports	O
Siri	B-Application
and	O
other	O
products	O
.	O
</s>
<s>
Bingbot	B-Protocol
is	O
the	O
name	O
of	O
Microsoft	O
's	O
Bing	B-Application
webcrawler	B-Application
.	O
</s>
<s>
It	O
replaced	O
Msnbot	B-Protocol
.	O
</s>
<s>
Baiduspider	O
is	O
Baidu	B-Application
's	O
web	B-Application
crawler	I-Application
.	O
</s>
<s>
Googlebot	B-Application
is	O
described	O
in	O
some	O
detail	O
,	O
but	O
the	O
reference	O
is	O
only	O
about	O
an	O
early	O
version	O
of	O
its	O
architecture	O
,	O
which	O
was	O
written	O
in	O
C++	O
and	O
Python	B-Language
.	O
</s>
<s>
WebCrawler	B-Application
was	O
used	O
to	O
build	O
the	O
first	O
publicly	O
available	O
full-text	B-Application
index	I-Application
of	O
a	O
subset	O
of	O
the	O
Web	O
.	O
</s>
<s>
It	O
was	O
based	O
on	O
lib-WWW	B-Protocol
to	O
download	O
pages	O
,	O
and	O
another	O
program	O
to	O
parse	O
and	O
order	O
URLs	O
for	O
breadth-first	B-Algorithm
exploration	O
of	O
the	O
Web	B-Application
graph	I-Application
.	O
</s>
<s>
WebFountain	B-Application
is	O
a	O
distributed	O
,	O
modular	O
crawler	O
similar	O
to	O
Mercator	O
but	O
written	O
in	O
C++	O
.	O
</s>
<s>
Xenon	B-Protocol
is	O
a	O
web	B-Application
crawler	I-Application
used	O
by	O
government	O
tax	O
authorities	O
to	O
detect	O
fraud	O
.	O
</s>
<s>
The	O
following	O
web	B-Application
crawlers	I-Application
are	O
available	O
,	O
for	O
a	O
price::	O
</s>
<s>
GNU	B-Protocol
Wget	I-Protocol
is	O
a	O
command-line-operated	O
crawler	O
written	O
in	O
C	B-Language
and	O
released	O
under	O
the	O
GPL	B-License
.	O
</s>
<s>
GRUB	B-Application
was	O
an	O
open	O
source	O
distributed	O
search	O
crawler	O
that	O
Wikia	O
Search	O
used	O
to	O
crawl	O
the	O
web	O
.	O
</s>
<s>
Heritrix	B-Application
is	O
the	O
Internet	O
Archive	O
's	O
archival-quality	O
crawler	O
,	O
designed	O
for	O
archiving	O
periodic	O
snapshots	O
of	O
a	O
large	O
portion	O
of	O
the	O
Web	O
.	O
</s>
<s>
It	O
was	O
written	O
in	O
Java	B-Language
.	O
</s>
<s>
ht://Dig	B-Application
includes	O
a	O
Web	B-Application
crawler	I-Application
in	O
its	O
indexing	O
engine	O
.	O
</s>
<s>
HTTrack	B-Language
uses	O
a	O
Web	B-Application
crawler	I-Application
to	O
create	O
a	O
mirror	O
of	O
a	O
web	O
site	O
for	O
off-line	O
viewing	O
.	O
</s>
<s>
It	O
is	O
written	O
in	O
C	B-Language
and	O
released	O
under	O
the	O
GPL	B-License
.	O
</s>
<s>
Apache	B-Language
Nutch	I-Language
is	O
a	O
highly	O
extensible	O
and	O
scalable	O
web	B-Application
crawler	I-Application
written	O
in	O
Java	B-Language
and	O
released	O
under	O
an	O
Apache	B-Application
License	I-Application
.	O
</s>
<s>
It	O
is	O
based	O
on	O
Apache	B-Application
Hadoop	I-Application
and	O
can	O
be	O
used	O
with	O
Apache	B-Language
Solr	I-Language
or	O
Elasticsearch	B-Application
.	O
</s>
<s>
Open	B-Application
Search	I-Application
Server	I-Application
is	O
a	O
search	B-Application
engine	I-Application
and	O
web	B-Application
crawler	I-Application
software	O
release	O
under	O
the	O
GPL	B-License
.	O
</s>
<s>
Scrapy	B-Protocol
,	O
an	O
open	O
source	O
webcrawler	B-Application
framework	O
,	O
written	O
in	O
python	B-Language
(	O
licensed	O
under	O
BSD	B-Operating_System
)	O
.	O
</s>
<s>
Seeks	O
,	O
a	O
free	O
distributed	O
search	B-Application
engine	I-Application
(	O
licensed	O
under	O
AGPL	B-Application
)	O
.	O
</s>
<s>
StormCrawler	B-Language
,	O
a	O
collection	O
of	O
resources	O
for	O
building	O
low-latency	O
,	O
scalable	O
web	B-Application
crawlers	I-Application
on	O
Apache	B-General_Concept
Storm	I-General_Concept
(	O
Apache	B-Application
License	I-Application
)	O
.	O
</s>
<s>
tkWWW	B-Application
Robot	I-Application
,	O
a	O
crawler	O
based	O
on	O
the	O
tkWWW	B-Application
web	O
browser	O
(	O
licensed	O
under	O
GPL	B-License
)	O
.	O
</s>
<s>
Xapian	B-Application
,	O
a	O
search	O
crawler	O
engine	O
,	O
written	O
in	O
c++	O
.	O
</s>
<s>
YaCy	B-Application
,	O
a	O
free	O
distributed	O
search	B-Application
engine	I-Application
,	O
built	O
on	O
principles	O
of	O
peer-to-peer	O
networks	O
(	O
licensed	O
under	O
GPL	B-License
)	O
.	O
</s>
