李改, 徐清振, 李磊, 黄锦涛. TLRank:一种新的社会化协同排序推荐算法[J]. 华南师范大学学报(自然科学版), 2019, 51(5): 121-128. doi: 10.6054/j.jscnun.2019094
引用本文: 李改, 徐清振, 李磊, 黄锦涛. TLRank:一种新的社会化协同排序推荐算法[J]. 华南师范大学学报(自然科学版), 2019, 51(5): 121-128. doi: 10.6054/j.jscnun.2019094
LI Gai, XU Qingzhen, LI Lei, HUANG Jintao. TLRank:A New Social Collaborative Ranking Recommendation Algorithm[J]. Journal of South China Normal University (Natural Science Edition), 2019, 51(5): 121-128. doi: 10.6054/j.jscnun.2019094
Citation: LI Gai, XU Qingzhen, LI Lei, HUANG Jintao. TLRank:A New Social Collaborative Ranking Recommendation Algorithm[J]. Journal of South China Normal University (Natural Science Edition), 2019, 51(5): 121-128. doi: 10.6054/j.jscnun.2019094

TLRank:一种新的社会化协同排序推荐算法

TLRank:A New Social Collaborative Ranking Recommendation Algorithm

  • 摘要: 已有的社会化协同排序推荐算法的研究只是简单地融入用户的社交网络信息,没有考虑用户之间社会化信任网络的传递性;同时,该推荐算法的性能面临数据高度稀疏性问题的挑战.为了进一步解决这些问题,在传统的协同排序推荐算法(ListRank, List-wise Learning to Rank)和最新的社会化协同过滤算法(TrustMF, Social Collaborative Filtering by Trust)的基础上,提出了一种新的社会化协同排序推荐算法(TLRank),融合均高度稀疏的用户的显式评分数据和社会化信任网络数据,以进一步增强协同排序推荐算法的性能.实验结果表明:在各个评价指标下,TLRank算法的性能均优于几个经典的协同排序推荐算法,且复杂度低、运算时间与评分点个数线性相关;TLRank算法的推荐精度高、可扩展性好,适合处理大数据,可广泛运用于互联网信息推荐领域.

     

    Abstract: The previous research on social collaborative ranking (SCR) recommendation algorithm only integrates the user's social network information into their model, but does not take into account the transmission of social trust network between users. A problem of the previous research on SCR is data sparsity and cold start that severely degrades quality of recommendation. To solve the problem, a new social collaborative ranking recommendation algorithm (TLRank) based on ListRank (List-wise Learning to Rank) algorithm and the newest TrustMF (Social Collaborative Filtering by Trust) algorithm is proposed to improve the performance of collaborative ranking recommendation by means of elaborately integrating twofold sparse information, the conventional rating data given by users and the social network among the same users. Experimental results on practical datasets show that our proposed TLRank algorithm outperforms the existing CR approaches with reference to the different evaluation metrics and that the algorithm has the advantage of low complexity and is shown to be linear with the number of observed ratings in a given user-item rating matrix. Because of its high precision and good expansibility, TLRank is suitable for processing big data and has a prospect of wide application in the field of internet information recommendation.

     

/

返回文章
返回