基于物品的协同过滤算法---《推荐系统实践》---Python源码（10）

一、源码说明

基于物品的协同过滤算法和基于用户的协同过滤算法类似，以给该物品评分的用户作为物品的特征向量，从而计算物品之间的余弦相似度。以下代码根据点击打开链接修改而来，修改了计算相似度的函数和进行推荐的函数。推荐效果的准确度不到10%，基于用户的准确度在20%。

二、准确度不高的原因分析

从推荐的结果看，根据代码设定是要推荐TOP-10的一个列表，但是结果往往很多只有3，4个，并没有10个。原因是用户u评分了物品i，与物品i大部分相似的物品已经存在于用户u的评分物品中呢，那么就不能再次推荐。

三、python源码

import random as rd , math as mt, operator as op
"""
SplitData(data, M, k, seed) approximately split data into M-1 train data and 1 test data.
This function should be call M times on the conditon that K varies from 0 to M-1 
and keep seed constant.

"""
def SplitData(data, M, k, seed):
    test = []
    train = []
    rd.seed(seed)
    for user,item in data:
        if rd.randint(0,M) == k:#generate a uniform random number in [0,M]
            test.append([user,item])
        else:
            train.append([user,item])
    return train, test
"""
the data structure of train data, test data are express a dictionary like {key:set}.
key is user. set is a set of movies that the users have rated.
"""
def list2dic(listdata):
    dicdata = dict()
    for user,item in listdata:
        if user not in dicdata.keys():
            dicdata[user] = set()
            dicdata[user].add(item)
        else:
            dicdata[user].add(item)
    return dicdata
#
def createItemtrain(train):
    itemtrain = dict()
    for user, item in train:
        if item not in itemtrain.keys():
            itemtrain[item] = set()
            itemtrain[item].add(user)
        else:
            itemtrain[item].add(user)
    return itemtrain

def ItemSimilarity1(itemtrain):
    W = dict()
    for i in itemtrain.keys():
        W[i]=dict()
        for j in itemtrain.keys():
            if i==j:
                continue
            W[i][j] = len(itemtrain[i] & itemtrain[j])
            W[i][j] /= mt.sqrt(len(itemtrain[i]) * len(itemtrain[j]) * 1.0)
    return W

# top-N list for user u 
def Recommand(train, W, K, N):
    allrank = dict()
    for u in train.keys():
        for i in train[u]:
            rank = dict()
            for j, wij in sorted(W[i].items(), key = op.itemgetter(1), reverse = True)[0:K]:
                if j not in train[u]:
                    if j not in rank.keys():
                        rank[j] = wij;
                    else:
                        rank[j] += wij * 1
        ranklist = []        
        for  item, sims in sorted(rank.items(), key = op.itemgetter(1), reverse = True)[0:N]:
            ranklist.append(item)
            allrank[u] = ranklist      
    return allrank
# compute precision
def Precision(allrank, test, N):
    hit = 0
    all = 0
    for user in test.keys():
        tu = test[user]
        if user in allrank.keys():
            for item in allrank[user]:
                if item in tu:
                    hit+=1
            all += N
    return hit / (all*1.0)
"""
main function
"""
filestring = '/home/sysu-hgavin/文档/ml-1m/ratings.dat'
f = open(filestring, 'r')
data = [] 
#count = 0
while 1:
    line = f.readline()#read data 
    #count += 1
    if not line:
        break
    #if count > 1000:
    #    break
    line = line.split("::")[:2]
    line[0] = int (line[0])
    line[1] = int (line[1])
    data.append(line)
f.close()
M = 8
seed = 2
N = 10 #top-N
K = 10 #the number of neighbors
#for k in range(M-1):
train, test = SplitData(data, M, 1, seed)#generate train data and test data
itemtrain = createItemtrain(train)

train = list2dic(train)
test = list2dic(test)
W = ItemSimilarity1(itemtrain)
allrank = Recommand(train, W, K, N)
precision = Precision(allrank, test, N)
print (precision)