MinHash算法+实现

参考:

原理ppt:  http://wenku.baidu.com/view/089e85c42cc58bd63186bdfc.html 

求解实现算法: http://fuliang.iteye.com/blog/1025638​ 最后部分. 感谢原作者.   算法原理+数学证明见原文.

简单的实现python代码如下:

 1 import os
 2 import sys
 3 
 4 
 5 def hash_func_demo1(x):
 6     return x % 5
 7 
 8 def hash_func_demo2(x):
 9     return (2 * x  + 1) % 5
10 
11 
12 ### data:[C1, C2, C3, ... CM]; C1:[a1, a2, a3 ... an]. thus D: n * m
13 ### hash_funcs;[h1, h2, ..., hr]
14 ### return: r * m matrix
15 def min_hash(data, hash_funcs):
16     MAX = 100000000
17     M, N, R = len(data), len(data[0]), len(hash_funcs)
18 
19     rt = []
20     for i in range(0, R):
21         rt.append(map(lambda x : x, [MAX] * M  ))
22     
23     for r in range(0, N):
24         hashes = map(lambda x : x(r + 1), hash_funcs)
25         for col in range(0, M):
26             if data[col][r] == 0:
27                 continue            
28             for k in range(0, R):
29                 rt[col][k] = min(rt[col][k], hashes[k])
30 
31     return rt
32 
33 
34 if __name__ == "__main__":
35     data = [[1, 0, 1, 1, 0],
36             [0, 1, 1, 0, 1],
37             ]
38 
39     hash_funcs = [hash_func_demo1, hash_func_demo2]
40     rt =  min_hash(data, hash_funcs)
41     print rt