<s>
In	O
data	B-General_Concept
analysis	I-General_Concept
,	O
the	O
self-similarity	B-Application
matrix	I-Application
is	O
a	O
graphical	O
representation	O
of	O
similar	O
sequences	O
in	O
a	O
data	O
series	O
.	O
</s>
<s>
Similarity	O
can	O
be	O
explained	O
by	O
different	O
measures	O
,	O
like	O
spatial	O
distance	O
(	O
distance	O
matrix	O
)	O
,	O
correlation	O
,	O
or	O
comparison	O
of	O
local	O
histograms	B-Algorithm
or	O
spectral	O
properties	O
(	O
e.g.	O
</s>
<s>
A	O
similarity	O
plot	O
can	O
be	O
the	O
starting	O
point	O
for	O
dot	B-Application
plots	I-Application
or	O
recurrence	B-Application
plots	I-Application
.	O
</s>
<s>
To	O
construct	O
a	O
self-similarity	B-Application
matrix	I-Application
,	O
one	O
first	O
transforms	O
a	O
data	O
series	O
into	O
an	O
ordered	O
sequence	O
of	O
feature	B-Algorithm
vectors	I-Algorithm
,	O
where	O
each	O
vector	O
describes	O
the	O
relevant	O
features	O
of	O
a	O
data	O
series	O
in	O
a	O
given	O
local	O
interval	O
.	O
</s>
<s>
Then	O
similar	O
segments	O
of	O
feature	B-Algorithm
vectors	I-Algorithm
will	O
show	O
up	O
as	O
path	O
of	O
high	O
similarity	O
along	O
diagonals	O
of	O
the	O
matrix	O
.	O
</s>
<s>
and	O
for	O
audio	O
segmentation	O
using	O
spectral	B-Algorithm
clustering	I-Algorithm
of	O
the	O
self-similarity	B-Application
matrix	I-Application
.	O
</s>
