实现论文SoRec Social Recommendation Using Probabilistic Matrix Factorization

github地址
论文地址

数据集： Epinion dataset
数据网址：http://www.datatang.com/datares/go.aspx?dataid=602020
The dataset contains: 49,290 users who rated a total of 139,738 different items at least once,
writing 664,824 reviews. 487,181 issued trust statements.

Sorec.py

import math
import numpy as np
from numpy.random import RandomState
from scipy.sparse import csr_matrix
from scipy.sparse import coo_matrix
from matplotlib import pyplot as plt
import time
import copy


def fx(x):
    return (x - 1) / 4

def sigmoid(x):
    return 1 / (1 + np.exp(-x))

# The derivative of sigmoid
def sigmoid_(x):
    val = sigmoid(x)
    return val*(1-val)

class MF:
    def __init__(self, R, Validate, C, lr = 0.01,momentum=0.8,
                 lambda_c=10, lambda_u=0.001, lambda_v=0.001, lambda_z=0.001, latent_size=10,iters=1000,seed=None):
        """
        :param R: train set of ratings_data
        :param Validate: validate set of ratings_data
        :param C: trust matrix of users
        :param lr: learning rate
        :param momentum:
        :param lambda_c:
        :param lambda_u:
        :param lambda_v:
        :param lambda_z:
        :param latent_size:
        :param iters: iteration
        :param seed:
        """
        self.R = R
        self.Validate= Validate
        self.C = C
        self.lambda_c = lambda_c
        self.lambda_u = lambda_u
        self.lambda_z = lambda_z
        self.lambda_v = lambda_v
        self.latent_size = latent_size
        self.random_state = RandomState(seed)
        self.iters = iters
        self.lr = lr
        self.U = np.mat(self.random_state.rand(latent_size, np.size(R, 0)))
        self.V = np.mat(self.random_state.rand(latent_size, np.size(R, 1)))
        self.Z = np.mat(self.random_state.rand(latent_size, np.size(C, 1)))
        self.momentum = momentum


    # the MAE for train set
    def train_loss(self, UVdata):
        loss = (np.fabs(4 * sigmoid(UVdata) + 1 - self.R.data)).sum()
        loss /= self.R.shape[0]
        return loss

    # the MAE for validate_set
    def val