计算机工程 ›› 2019, Vol. 45 ›› Issue (12): 176-181,200.doi: 10.19678/j.issn.1000-3428.0052783

• 人工智能及识别技术 • 上一篇    下一篇

基于矩阵填充与改进PSO算法的多标准协同过滤

叶莉1, 吴春明1, 强保华2, 谢武3   

  1. 1. 西南大学 计算机与信息科学学院, 重庆 400715;
    2. 广西密码学与信息安全重点实验室, 广西 桂林 541004;
    3. 广西高校云计算与复杂系统重点实验室, 广西 桂林 541004
  • 收稿日期:2018-09-29 修回日期:2018-12-19 发布日期:2018-12-26
  • 作者简介:叶莉(1993-),女,硕士,主研方向为智能推荐系统;吴春明,副教授、博士;强保华、谢武,博士。
  • 基金项目:
    国家自然科学基金(61762025);西南大学自然科学基金(SWU112017);广西密码学与信息安全重点实验室项目(GCIS201709);广西云计算与大数据合作创新中心项目(YD16E04);广西高校云计算与复杂系统重点实验室项目(14106,15204)。

Multi-Criteria Collaborative Filtering Based on Matrix Filling and Improved PSO Algorithm

YE Li1, WU Chunming1, QIANG Baohua2, XIE Wu3   

  1. 1. College of Computer and Information Science, Southwest University, Chongqing 400715, China;
    2. Guangxi Key Laboratory of Cryptography and Information Security, Guilin, Guangxi 541004, China;
    3. Guangxi Colleges and Universities Key Laboratory of Cloud Computing and Complex System, Guilin, Guangxi 541004, China
  • Received:2018-09-29 Revised:2018-12-19 Published:2018-12-26

摘要: 在多标准协同过滤中,存在稀疏性处理方法单一以及传统粒子群优化(PSO)算法早熟、易陷入局部最优等问题。为此,基于矩阵填充及改进PSO算法,提出一种多标准协同过滤模型。采用矩阵填充方法对稀疏数据的缺失部分进行估算,以避免降维方法对原始数据信息造成损失,同时结合高斯算子快速收敛的优势以及遗传算子对生物进化模拟的有效性对PSO算法进行改进,聚合多标准评分生成TopN推荐列表。实验结果表明,与基于标准PSO算法以及基于遗传算子改进PSO算法的模型相比,该模型的评分预测准确度较优,能为个性化推荐提供有效的支持。

关键词: 多标准协同过滤, 矩阵填充, 改进粒子群优化算法, 高斯算子, 遗传算子

Abstract: In multi-criteria collaborative filtering,current methods for sparsity processing are relatively insufficient,and the traditional Particle Swarm Optimization(PSO) algorithm has the drawbacks of premature convergence and easy to fall into local optimum.So,we propose a multi-criteria collaborative filtering model based on matrix filling and improved PSO algorithm.First,we use matrix filling method to estimate the missing part of the sparse data,so as to avoid the loss of the original data information caused by the dimensionality reduction method.Then,we present an improved PSO algorithm,which fully combines the advantages of fast convergence of Gaussian operators and the effectiveness of genetic operators on biological evolution simulation.Finally,a TopN recommendation list is generated through aggregated multi-criteria score.Experimental results show that compared with the P_MCCF model and the GIP_MCCF model,the proposed model has better scoring prediction accuracy,which can provide effective support for personalized recommendations.

Key words: multi-criteria collaborative filtering, matrix filling, improved Particle Swarm Optimization(PSO) algorithm, Gaussian operator, genetic operator

中图分类号: