<s>
The	O
greedy	B-Algorithm
randomized	I-Algorithm
adaptive	I-Algorithm
search	I-Algorithm
procedure	I-Algorithm
(	O
also	O
known	O
as	O
GRASP	O
)	O
is	O
a	O
metaheuristic	B-Algorithm
algorithm	O
commonly	O
applied	O
to	O
combinatorial	O
optimization	O
problems	O
.	O
</s>
<s>
GRASP	O
typically	O
consists	O
of	O
iterations	O
made	O
up	O
from	O
successive	O
constructions	O
of	O
a	O
greedy	B-Algorithm
randomized	B-General_Concept
solution	O
and	O
subsequent	O
iterative	O
improvements	O
of	O
it	O
through	O
a	O
local	B-Algorithm
search	I-Algorithm
.	O
</s>
<s>
The	O
greedy	B-Algorithm
randomized	B-General_Concept
solutions	O
are	O
generated	O
by	O
adding	O
elements	O
to	O
the	O
problem	O
's	O
solution	O
set	O
from	O
a	O
list	O
of	O
elements	O
ranked	O
by	O
a	O
greedy	B-Algorithm
function	O
according	O
to	O
the	O
quality	O
of	O
the	O
solution	O
they	O
will	O
achieve	O
.	O
</s>
<s>
To	O
obtain	O
variability	O
in	O
the	O
candidate	O
set	O
of	O
greedy	B-Algorithm
solutions	O
,	O
well-ranked	O
candidate	O
elements	O
are	O
often	O
placed	O
in	O
a	O
restricted	O
candidate	O
list	O
(	O
RCL	O
)	O
,	O
and	O
chosen	O
at	O
random	O
when	O
building	O
up	O
the	O
solution	O
.	O
</s>
<s>
This	O
kind	O
of	O
greedy	B-Algorithm
randomized	B-General_Concept
construction	O
method	O
is	O
also	O
known	O
as	O
a	O
semi-greedy	O
heuristic	O
,	O
first	O
described	O
in	O
Hart	O
and	O
Shogan	O
(	O
1987	O
)	O
.	O
</s>
<s>
There	O
are	O
also	O
techniques	O
for	O
search	O
speed-up	O
,	O
such	O
as	O
cost	O
perturbations	O
,	O
bias	O
functions	O
,	O
memorization	O
and	O
learning	O
,	O
and	O
local	B-Algorithm
search	I-Algorithm
on	O
partially	O
constructed	O
solutions	O
.	O
</s>
