-
Notifications
You must be signed in to change notification settings - Fork 3
/
sac1.py
128 lines (111 loc) · 4.04 KB
/
sac1.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
from igraph import *
from scipy import spatial
def qnewman(g, x, communityvertices):
m = g.ecount()
c = 1.0 / (2 * m)
G = 0
for i in communityvertices:
if g.are_connected(i, x):
G += g.es[g.get_eid(i, x)]["weight"]
dx = sum([g.es[e]["weight"] for e in g.incident(x)])
di = 0
for v in communityvertices:
di += sum([g.es[e]["weight"] for e in g.incident(v)])
return c * (G - (dx * di * c))
def main():
g = Graph()
if len(sys.argv) != 2:
print "usage : python sac1.py <alpha(0|0.5|1)>"
exit(1)
alpha = float(sys.argv[1])
with open("data/fb_caltech_small_attrlist.csv") as f:
attributes = next(f).strip().split(',')
for line in f:
attrs = map(float, line.strip().split(','))
g.add_vertex(**dict(zip(attributes, attrs)))
with open("data/fb_caltech_small_edgelist.txt") as f:
for line in f:
vertices = map(int, line.strip().split(' '))
g.add_edge(vertices[0], vertices[1], **{"weight": 1.0})
iteration = 0
finalans = range(g.vcount())
while iteration < 15:
iteration += 1
print "Iteration : ", iteration
communities, community = phase1(g, alpha)
phase2(g, community, communities, finalans)
if len(community) == g.vcount():
break
finalcommunities = [[] for i in xrange(g.vcount())]
for i in xrange(len(finalans)):
finalcommunities[finalans[i]].append(i)
print finalcommunities
if alpha == 0:
name = '0'
elif alpha == 1:
name = '1'
else:
name = '5'
with open("communities_" + name + ".txt", "w") as f:
for c in finalcommunities:
f.write(','.join(map(str, c)))
f.write('\n')
def phase2(g, community, communities, finalans):
d = dict(enumerate(communities.keys()))
d = {v: k for k, v in d.items()}
finalans[:] = [d[community[x]] for x in finalans]
community[:] = [d[x] for x in community]
g.contract_vertices(community, combine_attrs=mean)
g.simplify(combine_edges=sum)
def phase1(g, alpha):
community = range(g.vcount())
communities = {}
for i in community:
communities[i] = {i}
cosinesim = [[0 for x in xrange(g.vcount())] for y in xrange(g.vcount())]
for i in xrange(g.vcount()):
for j in xrange(i, g.vcount()):
cosinesim[i][j] = 1 - spatial.distance.cosine(g.vs[i].attributes().values(),
g.vs[j].attributes().values())
cosinesim[j][i] = cosinesim[i][j]
flag = True
loop = 0
while loop < 15 and flag:
flag = False
loop += 1
print "\tLoop : ", loop
for i in xrange(g.vcount()):
maxgain = 0
maxj = -1
original = community[i]
communities[original].discard(i)
prevcos = []
for x in communities[original]:
prevcos.append(cosinesim[i][x])
prevcosine = mean(prevcos)
communities[original].add(i)
for j in communities.keys():
if original != j:
cossim = []
for x in communities[j]:
cossim.append(cosinesim[i][x])
community[i] = j
compositemodularitygain = alpha * qnewman(g, i, communities[j]) + (1 - alpha) * (mean(
cossim) - prevcosine)
if compositemodularitygain > maxgain:
maxgain = compositemodularitygain
maxj = j
if maxj != -1:
flag = True
communities[original].discard(i)
communities[maxj].add(i)
community[i] = maxj
else:
community[i] = original
for c in communities.keys():
if not communities[c]:
del communities[c]
print "\t\tTotal : ", len(communities)
return communities, community
if __name__ == "__main__":
main()