simhash

#-*- coding: utf-8 -*-

import jieba.analyse
import numpy as np

def word_hash(t, w):
    if int(t)==1:
        return w
    else:
        return -w
def content_hash(t):
    if float(t)<=0:
        return ‘0‘
    else:
        return ‘1‘


def simhash(content):

    tags = jieba.analyse.extract_tags(content, topK=100, withWeight=True)

    hash_init = np.zeros(64)

    for t, w in tags:
        print t, w

        hash_str = str(bin(hash(t)).replace(‘0b‘, ‘‘).replace(‘-‘, ‘‘).zfill(64)[-64:])

        hash_str_deal = map(lambda x: word_hash(x, w), hash_str)

        hash_init =  hash_str_deal + hash_init
    return ‘‘.join(map(lambda x: content_hash(x), hash_init))

def similarity(hash1, hash2):
    pass
======================================================================================================================================================================

#-*- coding: utf-8 -*-

from simhash import simhash

if __name__ == ‘__main__‘:
    content1 = ‘而在感怀的热潮过后，关于人人网以及人人直播的价值讨论，开始浮出水面。‘
    content2 = ‘陈一舟还认为，相比年轻创业者，老江湖更适合做产业互联网。‘

    hash1 = simhash(content1)
    hash2 = simhash(content2)

    print len(filter(lambda x: x[0]==x[1], zip(hash1, hash2)))/64.0

simhash

原文：https://www.cnblogs.com/kayy/p/9967494.html