PageRank算法 python单机实现

海量数据挖掘课的编程作业。

实现PageRank，计算某个网页最终的rank值，数据是谷歌提供的。

作业反馈结果显示，代码正确。一共进行了26次迭代，总共运行时间83s。

数据链接http://snap.stanford.edu/data/web-Google.txt.gz。

代码：

from math import fabs
from time import time

data = open('web-Google.txt')
N =  875713
tax_rate = 0.8
eps = 1e-6
r = [1./N for i in range(N)]
r2 = [1./N for i in range(N)]
out_degree = [0 for i in range(N)]
m = [[] for i in range(N*2)]
hash_table = [-1 for i in range(N*2)]
idx = 0

def hash(x):
	global idx
	if hash_table[x] == -1:
		hash_table[x] = idx
		idx += 1
	return hash_table[x]

data.readline()
data.readline()
data.readline()
data.readline()
for line in data:
	x, y = map(hash, map(int, line.split()))
	out_degree[x] += 1
	m[y].append(x)

print 'data loaded'
print 'start iterating...'

t = 0
begin = time()

while True:
	for i in range(N):
		r[i] = 0
		for in_id in m[i]: 
			r[i] += tax_rate * r2[in_id] / out_degree[in_id]
	der = 1 - sum(r)
	for i in range(N):
		r[i] += der / N

	tag = 0
	for i in range(N):
		if fabs(r[i]-r2[i]) > eps:
			tag = 1
			break
	for i in range(N):
		r2[i] = r[i]
	t += 1
	if tag == 0:
		break

end = time()
print r[hash(99)]
print 'total iteration is %d' % t
print 'total time is %f' % (end - begin)