基于门控循环单元与主动学习的协同过滤推荐算法  

GRU-based collaborative filtering recommendation algorithm with active learning

作　　者：陈德蕾 王成[1] 陈建伟[2] 吴以茵[1] CHEN Delei;WANG Cheng;CHEN Jianwei;WU Yiyin(College of Computer Science and Technology,Huaqiao Universiiy,Xiamen 361021,Fujian,China;Department of Mathe matics and Statistics,San Diego State Universiiy,San Diego 92182,CA,USA)

机构地区：[1]华侨大学计算机科学与技术学院,福建厦门361021 [2]圣地亚哥州立大学数学与统计学院,加利福尼亚州圣地亚哥92182

出　　处：《山东大学学报（工学版）》2020年第1期21-27,48,共8页Journal of Shandong University（Engineering Science）

基　　金：福建省引导性科技计划资助项目(2017H01010065)。

摘　　要：针对传统协同过滤推荐算法存在无法反映用户短时兴趣的问题,提出一种基于门控循环单元(gated recurrent unit,GRU)神经网络与主动学习的协同过滤推荐算法。在采用GRU神经网络的基础上,将数据进行时序化处理,反映用户兴趣变化,并利用主动学习动态采样数据中的高质量的数据进行GRU神经网络的训练,使模型快速建立。在MovieLens1M数据集上的试验结果表明:加入主动学习的GRU模型的推荐算法比基于用户的协同过滤推荐算法(user-based collaborative filtering,UCF)、基于马尔科夫模型的协同过滤推荐算法(markov chain,MC)、基于隐语义模型的协同过滤推荐算法(latent factor model,LFM)算法有更高的短时预测率、召回率、项目覆盖率以及用户覆盖数,能够有效预测用户短时兴趣,提升精度,发掘长尾物品,且与原始GRU模型相比能够以更少的迭代次数达到相同效果。The traditional collaborative filtering recommendation algorithm failed to reflect short-term user interest.In order to reflect the short-term interests of users better,a collaborative filtering recommendation algorithm based on Gated Recurrent Unit(GRU)neural netw ork w ith active learning w as proposed.Based on the GRU neural netw ork,the algorithm processed the data into time-series data to reflect the change of the user s interest and used active learning to sample high-quality data dynamically for accelerating the training of GRU neural netw ork.The result on M ovieLens1 M dataset show ed that the GRU model w ith active learning could obtain higher short-term prediction success rate,recall rate,item coverage,and user coverage compared w ith the user-based collaborative filtering method(UCF),the markovian chain model-based collaborative filtering method(M C)and the matrix factory-based collaborative filtering method(LFM),so it could effectively predict the short-term interest of users,improve the accuracy,discover the long-tail items.M eanw hile,it could achieve the same effect w ith few er iterations compared w ith the original GRU model.

关 键 词：协同过滤 门控循环单元 主动学习 深度学习 时序化数据 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]