余弦相似度计算Python,java代码实现（可运行）

余弦相似度计算原理：

余弦相似度Python代码：

C保存的是分子部分的数据。N保存的是分母部分的数据。如果下面代码看起来比较费劲，可以看java代码，可能更容易理解一些。

## ItemCF-余弦算法
import math


def ItemSimilarity_cos(train):
    C = dict()   ##书本对同时被购买的次数
    N = dict()   ##书本被购买用户数
    for u,items in train.items():
        for i in items.keys():
            if i not in N.keys():
                N[i]=0
            N[i] += items[i]* items[i]
            for j in items.keys():
                if i == j:
                    continue
                if i not in C.keys():
                    C[i]=dict()
                if j not in C[i].keys():
                    C[i][j]=0
                ##当用户同时购买了i和j，则加评分乘积
                C[i][j] += items[i]*items[j]
    W = dict()  ##书本对相似分数
    for i,related_items in C.items():
        if i not in W.keys():
            W[i]=dict()
        for j,cij in related_items.items():
            W[i][j] = cij / (math.sqrt( N[i]) *math.sqrt( N[j]) )
    return W

if __name__ == '__main__':
    Train_Data = {'A':{'i1':1,'i2':1 ,'i4':1},
     'B':{'i1':1,'i4':1},
     'C':{'i1':1,'i2':1,'i5':1},
     'D':{'i2':1,'i3':1},
     'E':{'i3':1,'i5':1},
     'F':{'i2':1,'i4':1}
        }
    W= ItemSimilarity_cos (Train_Data)
    print(W)

结果：

{'i1': {'i2': 0.5773502691896258, 'i4': 0.6666666666666667, 'i5': 0.40824829046386296}, 'i2': {'i1': 0.5773502691896258, 'i4': 0.5773502691896258, 'i5': 0.35355339059327373, 'i3': 0.35355339059327373}, 'i4': {'i1': 0.6666666666666667, 'i2': 0.5773502691896258}, 'i5': {'i1': 0.40824829046386296, 'i2': 0.35355339059327373, 'i3': 0.4999999999999999}, 'i3': {'i2': 0.35355339059327373, 'i5': 0.4999999999999999}}

改代码来源自《推荐系统与深度学习》黄昕，赵伟，王本友——编著

因为实验需要，我又参照上述python代码，给出了Java版本，用Hashmap实现。代码如下
C的结构为<用户ID,<电影ID,评分>>
N的结构为<电影ID,电影评分的平方>
W的结构为<电影ID,<电影ID,相似度>>

Java代码

import java.util.HashMap;
import java.util.Map;

public class ItemSimilarity_cos {
    public static HashMap<Integer, HashMap<Integer, Float>> C; 
    public static HashMap<Integer, Float> N;   
    public static HashMap<Integer, HashMap<Integer, Float>> W;

    public static HashMap<Integer, HashMap<Integer, Float>> getC() {
        return C;
    }

    public static HashMap<Integer, Float> getN() {
        return N;
    }

    public static HashMap<Integer, HashMap<Integer, Float>> getW() {
        return W;
    }

    public ItemSimilarity_cos(HashMap<Integer, HashMap<Integer, Float>> map) {
        this.jisuan(map);
    }

    public static HashMap<Integer, HashMap<Integer, Float>> jisuan(HashMap<Integer, HashMap<Integer, Float>> map) {//map的输入是用户ID，电影ID和电影评分
        C = new HashMap<Integer, HashMap<Integer, Float>>();
        N = new HashMap<Integer, Float>();
        W = new HashMap<Integer, HashMap<Integer, Float>>();
        for (Map.Entry<Integer, HashMap<Integer, Float>> entry : map.entrySet()) {
            for (Map.Entry<Integer, Float> entry1 : entry.getValue().entrySet()) {
                if (!N.containsKey(entry1.getKey())) {
                    N.put(entry1.getKey(), 0F);

                }
                Float rate = N.get(entry1.getKey());
                rate += entry1.getValue()*entry1.getValue();
               // System.out.println(entry1.getKey()+"----------"+rate);
                N.put(entry1.getKey(), rate);

                for (Map.Entry<Integer, Float> entry2 : entry.getValue().entrySet()) {
                    if (entry1.getKey() == entry2.getKey()) {
                        continue;
                    }
                    if (!C.containsKey(entry1.getKey())) {
                        C.put(entry1.getKey(), new HashMap<Integer, Float>());
                    }
                    if (!C.get(entry1.getKey()).containsKey(entry2.getKey())) {
                        C.get(entry1.getKey()).put(entry2.getKey(),0F);
                    }
                    Float result= C.get(entry1.getKey()).get(entry2.getKey());
                    result += entry1.getValue() * entry2.getValue();
                    C.get(entry1.getKey()).put(entry2.getKey(),result);
                }
            }
        }
        for (Map.Entry<Integer, HashMap<Integer, Float>> entry : C.entrySet()) {
            if (!W.containsKey(entry.getKey())) {
                W.put(entry.getKey(), new HashMap<Integer, Float>());
            }
            for (Map.Entry<Integer, Float> entry1 : entry.getValue().entrySet()) {
                double result = entry1.getValue() / (Math.sqrt(N.get(entry.getKey()))* Math.sqrt(N.get(entry1.getKey())));
                W.get(entry.getKey()).put(entry1.getKey(), (float) result);
             //   System.out.println(result);
            }
        }
        return W;
    }

    public static void main(String[] args) {
        HashMap<Integer, HashMap<Integer, Float>> map = new HashMap<Integer, HashMap<Integer, Float>>();
        map.put(0, new HashMap<Integer, Float>());
        map.put(1, new HashMap<Integer, Float>());
        map.put(2, new HashMap<Integer, Float>());//用户ID
        map.get(0).put(1, 9.5F);
        map.get(0).put(2, 3.5F);
        map.get(0).put(3, 8.5F);
        map.get(0).put(4, 7.0F);
        map.get(1).put(1, 9.0F);
        map.get(1).put(2, 5.0F);
        map.get(2).put(1, 3.5F);
        map.get(2).put(2, 6.0F);
        map.get(2).put(4, 8.0F);
        ItemSimilarity_cos cos=new ItemSimilarity_cos(map);
       // System.out.println(cos.getW());
       // System.out.println(cos.getC());

    }
}

结果：

{1={2=0.8560688, 3=0.7013028, 4=0.6562579}, 2={1=0.8560688, 3=0.40894437, 4=0.79688376}, 3={1=0.7013028, 2=0.40894437, 4=0.6585046}, 4={1=0.6562579, 2=0.79688376, 3=0.6585046}}