基于评分预测与排序预测的协同过滤推荐算法

发布时间：2018-07-09 18:55

本文选题：推荐系统 + 协同排序　；参考：《电子学报》2017年12期

【摘要】：协同过滤推荐算法在电子商务领域运用广泛.之前的研究要么仅从评分预测的角度来研究,要么仅从排序预测的角度来研究.为了兼顾这两个方面,本文在传统的基于评分预测的PMF(Probabilistic Matrix Factorization)算法和基于排序预测的xCLiMF(Extended Collaborative Less-is-More Filtering)算法的基础上提出了一种基于评分预测与排序预测的协同过滤推荐算法URA(Unified Recommendation Algorithm),该方法通过在PMF和xCLiMF算法中共享用户和推荐对象的特征空间,利用PMF算法来学习高精度的用户和推荐对象的特征向量,从而进一步增强排序推荐性能.实验验证,该方法在评价指标NDCG和ERR下均优于PMF和xCLiMF算法,且复杂度与评分点个数线性相关.URA算法可运用于互联网信息推荐领域的大数据处理.
[Abstract]:Collaborative filtering recommendation algorithm is widely used in the field of electronic commerce. Previous studies have been conducted either from the perspective of score prediction or only from the perspective of ranking prediction. In order to balance these two aspects, Based on the traditional PMF (probabilistic Matrix factorization) algorithm based on score prediction and xCLiMF (extended Collaborative Less-is-More filtering) algorithm based on ranking prediction, this paper proposes a collaborative filtering recommendation algorithm URA (Unified recommendation Algorithm),) based on grading prediction and ranking prediction. In PMF and xCLiMF algorithms, the feature spaces of users and recommendation objects are shared, The PMF algorithm is used to study the feature vectors of high precision users and recommendation objects, so as to further enhance the performance of sorting recommendation. The experimental results show that the proposed method is superior to the PMF and xCLiMF algorithms under the evaluation indexes NDCG and err, and the complexity is linearly related to the number of scoring points. URA algorithm can be applied to the big data processing in the field of Internet information recommendation.
【作者单位】：顺德职业技术学院电子与信息工程学院;广东第二师范学院计算机科学系;中山大学数据科学与计算机学院;
【基金】：国家自然科学基金(No.61370186,No.61640222) 广东省自然科学基金项目(No.2016A030310018) 广东省科技计划项目(No.2014A010103040,No.2014B010116001) 广州市科技计划项目(No.201604010049,No.201510010203) 广东第二师范学院教授博士科研专项(No.2015ARF25)
【分类号】：TP391.3

【相似文献】