<s>
The	O
Canterbury	B-General_Concept
corpus	I-General_Concept
is	O
a	O
collection	O
of	O
files	B-Operating_System
intended	O
for	O
use	O
as	O
a	O
benchmark	O
for	O
testing	O
lossless	B-Algorithm
data	I-Algorithm
compression	I-Algorithm
algorithms	O
.	O
</s>
<s>
It	O
was	O
created	O
in	O
1997	O
at	O
the	O
University	O
of	O
Canterbury	O
,	O
New	O
Zealand	O
and	O
designed	O
to	O
replace	O
the	O
Calgary	B-General_Concept
corpus	I-General_Concept
.	O
</s>
<s>
The	O
files	B-Operating_System
were	O
selected	O
based	O
on	O
their	O
ability	O
to	O
provide	O
representative	O
performance	O
results	O
.	O
</s>
<s>
In	O
its	O
most	O
commonly	O
used	O
form	O
,	O
the	O
corpus	O
consists	O
of	O
11	O
files	B-Operating_System
,	O
selected	O
as	O
"	O
average	O
"	O
documents	O
from	O
11	O
classes	O
of	O
documents	O
,	O
totaling	O
2,810,784	O
bytes	O
as	O
follows	O
.	O
</s>
<s>
Additional	O
files	B-Operating_System
may	O
be	O
added	O
,	O
so	O
results	O
should	O
be	O
only	O
reported	O
for	O
individual	O
files	B-Operating_System
.	O
</s>
<s>
The	O
Artificial	O
Corpus	O
,	O
a	O
set	O
of	O
files	B-Operating_System
with	O
highly	O
"	O
artificial	O
"	O
data	O
designed	O
to	O
evoke	O
pathological	O
or	O
worst-case	O
bahavior	O
.	O
</s>
<s>
The	O
Large	O
Corpus	O
,	O
a	O
set	O
of	O
large	O
(	O
megabyte-size	O
)	O
files	B-Operating_System
.	O
</s>
