Documente Academic
Documente Profesional
Documente Cultură
substructure pattern
mining
Authors: Xifeng Yan and Jiawei Han
Copyright note:
Outlines
Background
Problem Definition
Authors Contribution
Experimental Result
Conclusion
3
Background
Background
G1=(V1,E1,L1)
1
2
4
(a)
G2=(V2,E2,L2)
V
W
U
(b)
f(V1.1)
f(V1.2)
f(V1.3)
f(V1.4)
f(V1.5)
=
=
=
=
=
V2.2
V2.5
V2.3
V2.4
V2.1
5
(c)
Two Isomorphic graph (a) and (b) with their mapping function (c)
Two graphs are isomorphic if one can find a mapping of nodes of the
first graph to the second graph such that labels on nodes and edges
are preserved.
Xifeng Yan
Authors Contribution
Anti-Monotonicity of graph
frequency
...
2-edge
...
...
...
...
3-edge
...
...
...
12
DFS code
a
b
Y
b
X
c
Z
v0
v1
v2
v3
v4
Y
b
X
c
Z
e0: (0,1,x,y,a)
a
e1: (1,2,y,x,b)
e2: (2,0,x,x,a)
e3: (2,3,x,z,c)
e4: (3,1,x,y,b)
e5: (1,4,x,z,d)
15
(B)
(C)
(D)
(0,1,x,y,a)
(0,1,y,x,a) (0,1,x,x,a)
(1,2,y,x,b)
(1,2,x,x,a) (1,2,x,y,b)
(2,0,x,x,a)
(2,0,x,y,b) (0,1,y,x,a)
(2,3,x,z,c)
(2,3,x,z,c) (2,3,y,z,a)
(3,1,z,y,b)
(3,0,z,y,b) (3,1,z,x,c)
(1,4,x,z,d)
(0,4,y,z,d) (2,4,y,z,d)
16
17
...
2-edge
...
...
...
...
3-edge
...
...
...
18
Theorem
to G1 iff min_dfs_code(G0)=min_dfs_code(G1).
Algorithm
20
Algorithm
21
Experimental Result
22
Experimental Result
23
Conclusion
24
Questions?
25
Exam Questions
Answer:
Unlike Apriori-based algorithms, gSpan does not generate candidate patterns and
tests for false positive pruning. This feature of gSpan is both time and space
efficient. Apriori-based algorithms must generate a candidate and then test for
isomorphism against graph dataset to calculate support. This test is costly. On
the other hand, gSpan does not test for isomorphism!
Answer:
27