IteRank: An iterative network-oriented approach to neighbor-based collaborative ranking
作者：
Highlights：
• IteRank is a novel framework for neighbor-based collaborative ranking.
• IteRank uses two new networks, UPNet and PRNet, to model and infer users’ priorities.
• IteRank exploits PRNet to calculate the users’ similarities and preferences’ concordance.
• IteRank refines the concordance and rank values by a random walk process on PRNet.
• IteRank significantly outperforms state-of-the-art collaborative ranking methods.
摘要
•IteRank is a novel framework for neighbor-based collaborative ranking.•IteRank uses two new networks, UPNet and PRNet, to model and infer users’ priorities.•IteRank exploits PRNet to calculate the users’ similarities and preferences’ concordance.•IteRank refines the concordance and rank values by a random walk process on PRNet.•IteRank significantly outperforms state-of-the-art collaborative ranking methods.
论文关键词：Collaborative ranking,Pairwise preference,Graph-based recommendation,Heterogeneous network,Ranking similarity,Iterative refinement
论文评审过程：Received 7 October 2016, Revised 16 March 2017, Accepted 3 May 2017, Available online 4 May 2017, Version of Record 25 May 2017.
论文官网地址：https://doi.org/10.1016/j.knosys.2017.05.002
