TrustOCCR:基于信任的社会化单类协同排序推荐算法
TrustOCCR:Social One Class Collaborative Ranking Recommendation Algorithm by Trust
李改 1郭泽浩1
作者信息
- 1. 顺德职业技术学院智能制造学院,广东 佛山 528333
- 折叠
摘要
之前关于社会化单类协同排序推荐算法的研究仅仅集成用户的社交网络信息到推荐模型中,未能考虑到用户的社交信任网络的传递性.为了解决该问题,基于最新的CLiMF模型和TrustMF模型,提出了一种新的基于信任的社会化单类协同排序推荐算法(TrustOCCR).该算法通过精心整合双重稀疏信息(隐式评分矩阵和具有传递性的社交信任网络矩阵)来进一步提高社会化单类协同排序推荐算法的性能.在真实的实验数据集上验证,采用两个不同的评价指标,本文提出的TrustOCCR算法均优于最新的社会化单类协同排序推荐算法.且TrustOCCR算法可扩展性好,适合在互联网信息推荐领域用于处理大数据.
Abstract
The problem with previous studies of social One Class Collaborative Ranking(OCCR)algorithms is that they simply integrated the user's social network information into their model,without taking into account the transmission of social trust networks between users.To solve this problem,a new social one class collaborative ranking recommendation algorithm(TrustOCCR)based on CLiMF model and the newest TrustMF model is proposed,which aims to improve the performance of social one class collaborative ranking recommendation by integrating twofold sparse data,i.e.,implicit feedback data and the transitive social trust network data.Experimental results on practical dataset show that our proposed TrustOCCR algorithm outperformed existing state-of-the-art OCCF approach over different evaluation metrics,and that the TrustOCCR algorithm possesses good expansibility and is suitable for processing big data in the field of internet information recommendation.
关键词
推荐系统/协同过滤/社会化协同排序/社交网络Key words
recommendation system/collaborative filtering/social collaborative ranking/social network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据