矩阵分解技术在电影推荐系统中的应用  

Application of Matrix Factorization in Movie Recommendation System

作　　者：蔡崇超 许华虎[1] CAI Chong-chao;XU Hua-hu(School of Computer Engineering and Science,Shanghai University,Shanghai 200444,China;College of Logistic and Information Engineering,Huzhou Vocational&Technical College,Huzhou 313000,China)

机构地区：[1]上海大学计算机工程与科学学院,上海200444 [2]湖州职业技术学院物流与信息工程学院,浙江湖州313000

出　　处：《软件导刊》2021年第1期174-177,共4页Software Guide

摘　　要：随着大数据、移动互联网的快速发展,推荐系统成为解决网络信息过载的有力工具。为解决传统推荐系统由于没有将社交网络中用户关系考虑进去而导致的稀疏矩阵、冷启动等问题,提出一种基于矩阵分解技术的电影推荐系统算法MFMRS。该算法充分考虑到社交网络中用户之间的关系对推荐结果的影响,通过设置特征参数、损失函数、随机梯度下降等方法对推荐系统的精度进行改进。结果表明,通过应用该算法,Douban数据集的精度提升62%,Netflix数据集的精度提升51%。With the rapid development of big data,recommendation system has become a powerful tool to solve network information overload.The main purpose of this paper is to solve the problem of sparse matrix and cold start in traditional recommendation system,which does not take the user relationship into account.In this paper,an algorithm of movie recommendation system based on matrix fac⁃torization is proposed.The algorithm fully considers the influence of the relationship between users in social network on the recommen⁃dation results,and solves the above problems by setting feature parameters,gradient descent and offset variables.The results show that the accuracy of Doublan dataset and Netflix dataset is improved by 62%and 51%respectively by using this algorithm.

关 键 词：推荐系统 社交网络 矩阵分解 梯度下降 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]