<s>
The	O
MAtrixware	O
REsearch	O
Collection	O
(	O
MAREC	B-General_Concept
)	O
is	O
a	O
standardised	O
patent	O
data	O
corpus	O
available	O
for	O
research	O
purposes	O
.	O
</s>
<s>
MAREC	B-General_Concept
seeks	O
to	O
represent	O
patent	O
documents	O
of	O
several	O
languages	O
in	O
order	O
to	O
answer	O
specific	O
research	O
questions	O
.	O
</s>
<s>
It	O
consists	O
of	O
19	O
million	O
patent	O
documents	O
in	O
different	O
languages	O
,	O
normalised	O
to	O
a	O
highly	O
specific	O
XML	B-Protocol
schema	O
.	O
</s>
<s>
MAREC	B-General_Concept
is	O
intended	O
as	O
raw	O
material	O
for	O
research	O
in	O
areas	O
such	O
as	O
information	B-Library
retrieval	I-Library
,	O
natural	B-Language
language	I-Language
processing	I-Language
or	O
machine	B-Application
translation	I-Application
,	O
which	O
require	O
large	O
amounts	O
of	O
complex	O
documents	O
.	O
</s>
<s>
In	O
MAREC	B-General_Concept
,	O
the	O
documents	O
from	O
different	O
countries	O
and	O
sources	O
are	O
normalised	O
to	O
a	O
common	O
XML	B-Protocol
format	O
with	O
a	O
uniform	O
patent	O
numbering	O
scheme	O
and	O
citation	O
format	O
.	O
</s>
<s>
MAREC	B-General_Concept
is	O
a	O
comparable	O
corpus	O
,	O
where	O
many	O
documents	O
are	O
available	O
in	O
similar	O
versions	O
in	O
other	O
languages	O
.	O
</s>
<s>
The	O
19,386,697	O
XML	B-Protocol
files	O
measure	O
a	O
total	O
of	O
621	O
GB	O
and	O
are	O
hosted	O
by	O
the	O
Information	B-Library
Retrieval	I-Library
Facility	O
.	O
</s>
<s>
MAREC	B-General_Concept
is	O
used	O
in	O
the	O
Patent	O
Language	O
Translations	O
Online	O
(	O
PLuTO	O
)	O
project	O
.	O
</s>
