文章摘要
张大鹏*,张伟*,张丽君*,刘雅军**.基于综合信任的奇异值分解推荐算法研究[J].高技术通讯(中文),2021,31(1):102~112
基于综合信任的奇异值分解推荐算法研究
  
DOI:10.3772/j.issn.1002-0470.2021.01.011
中文关键词: 推荐系统; 信任网络; 综合信任; 奇异值分解(SVD)
英文关键词: recommender system, trust network, comprehensive trust, singular value decomposition (SVD)
基金项目:
作者单位
张大鹏*  
张伟*  
张丽君*  
刘雅军**  
摘要点击次数: 1243
全文下载次数: 907
中文摘要:
      针对传统矩阵分解推荐算法中数据稀疏、冷启动和用户信任矩阵数据稀疏等问题,本文提出了一种改进的基于路径的信任计算模型,利用用户-用户之间的直接信任关系和受信任者对用户信任关系的影响,计算用户-用户之间存在的间接信任关系,从而填充用户-用户信任矩阵。在此基础上,将用户之间的信任关系与奇异值分解(SVD)模型相结合,提出了一种融合综合信任的奇异值分解算法,即CT-SVD算法。该推荐算法结合用户评分矩阵和信任关系矩阵,对传统的奇异值分解推荐算法进行优化,提高了推荐系统评分预测的准确性。在FilmTrust和Ciao数据集上的实验结果表明,该算法能够有效地缓解推荐系统的数据稀疏性和冷启动问题。
英文摘要:
      Aiming at the problems of data sparseness, cold start and sparse data of user trust matrix in traditional matrix decomposition recommendation algorithm, an improved path based trust computing model is proposed, which uses the direct trust relationship between users-users and the influence of the trusted on the trust relationship between users-users to calculate the indirect trust relationship between users-users and fill in the users-users trust matrix. On this basis, combining the trust relationship between users and the singular value decomposition (SVD) model, a singular value decomposition algorithm combining synthetic trust is proposed. The recommendation algorithm combines the user scoring matrix and the trust relationship matrix to the traditional singular value. The decomposition recommendation algorithm is optimized to improve the accuracy of the recommendation system score prediction. Experimental results on the FilmTrust and Ciao datasets show that the proposed algorithm can effectively alleviate the data sparsity and cold start problems of the recommended system.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮