一种基于矩阵分块技巧的协同过滤跨域推荐  

A Collaborative Filtering Cross-Domain Recommendation Based on Matrix Blocking Technique

作　　者：姜树媛 胡建华[1] 王新利[1] 

机构地区：[1]上海理工大学理学院,上海

出　　处：《建模与仿真》2023年第3期2091-2101,共11页Modeling and Simulation

摘　　要：针对数据稀疏性的挑战和冷启动问题,跨域推荐技术成为研究热点,大大提高了推荐的精确度。但当前主要的跨域推荐模型建立在源域与目标域的用户或项目完全重叠的情形下,其应用具有局限性。本文拟在用户部分重叠的场景下,提出一种新的跨域推荐模型:基于矩阵分块技巧的协同过滤跨域推荐(A Collaborative Filtering Cross-Domain Recommendation Based on Matrix Blocking Technique, CFCDRMB)。新模型采用矩阵三分解形式,利用矩阵分块技巧表征共享用户潜在因子和域特有用户潜在因子,同时用联合矩阵分解学习特征矩阵来捕获各自域的数据特征并实现共享知识的迁移。本文在3个数据集上与4个方法进行了对比,实验结果表明,新提出的模型在部分用户重叠场景下具有显著的优越性。Due to the challenge of data sparsity and cold start, cross-domain recommendation technology has become a research hotspot, greatly improving the accuracy of recommendation. However, the cur-rent main cross-domain recommendation model is based on the situation that the users or items in the source domain and the target domain completely overlap, which leads to the limitation in ap-plication. In this paper, we have proposed a new cross-domain recommendation model under the scenario of partial overlap of users: A Collaborative Filtering Cross-Domain Recommendation based on Matrix Blocking Technique (CFCDRMB). The new model adopts the matrix triple-decomposition form, uses the matrix blocking technique to represent the shared user potential factor and the do-main-specific user potential factor, and utilizes the joint matrix decomposition learning feature matrix to capture the data characteristics of each domain and transfer the shared knowledge. In this paper, four methods are compared on three data sets, and the experimental results show that the new model has significant advantages in partial user overlap scenarios.

关 键 词：协同过滤 数据稀疏性 矩阵分块 推荐模型 数据特征 特征矩阵 潜在因子 跨域 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]