语句相似度

参考链接：https://zhuanlan.zhihu.com/p/88938220
https://blog.csdn.net/yjw123456/article/details/107923566
https://blog.csdn.net/betterzl/article/details/109983541
列表的交并差：https://blog.csdn.net/qdPython/article/details/118802922
jieba的使用：https://blog.csdn.net/lukabruce/article/details/82351742

import random,jieba

class Similarity():
    def __init__(self,a,b):
        # 分词
        self.a = jieba.lcut(a)
        self.b = jieba.lcut(b)
        # 词袋 a,b的并集
        self.word_bag = list(set(self.a).union(set(self.b)))
        print('词袋：',self.word_bag)
        # 词频
        self.wf1 = self.word_frequency(self.a)
        self.wf2 = self.word_frequency(self.b)
        print('a词频：',self.wf1)
        print('b词频：',self.wf2)

    # 统计词频
    def word_frequency(self,word):
        c = []
        for i in self.word_bag:
            if i in word:
                c.append(1)
            else:
                c.append(0)
        return c
    # 欧氏距离
    def euclidean_distance(self):
        # 统计
        count = 0
        for i, t in enumerate(self.word_bag):
            count += (self.wf1[i] - self.wf2[i]) ** 2
        print('欧氏距离：',1 / (count ** 0.5))
        return 1 / (count ** 0.5)
    # 余弦距离
    def cosine_distance(self):
        # 统计
        count1, count2, count3 = 0, 0, 0
        for i, t in enumerate(self.word_bag):
            count1 += self.wf1[i] * self.wf2[i]
            count2 += self.wf1[i] * self.wf1[i]
            count3 += self.wf2[i] * self.wf2[i]
        print('余弦距离：',count1 / ((count1 ** 0.5) * (count3 ** 0.5)))
        return count1 / ((count1 ** 0.5) * (count3 ** 0.5))
    # Jacard相似度
    def Jacard_distance(self):
        count = 0
        for i in self.a:
            if i in self.b:
                count += 1
        print('Jacard：',count / len(max(self.a, self.b)))
        return count / len(max(self.a, self.b))
    # 海明距离
    def hamming_distance(self):
        c = 0
        for i, t in enumerate(self.word_bag):
            if self.wf1[i] == self.wf2[i]:
                c += 1
            else:
                c += 0
        print('海明距离：',c/len(max(self.a, self.b)))
        return c