摘要
针对传统协同过滤必须存在近邻用户才能给出推荐的问题,提出一种基于深度神经网络的推荐算法。先建立项目类型与用户项目评分的对应关系用于DNN训练,再建立一种带有Dropout策略的多分类深度神经网络模型,最后经过模型训练和反编码得到目标项目的预测评分。实验表明,新算法可以根据用户的评论项目数给出合理的推荐。
To solve the problem that traditional collaborative filtering must exist near users to give recommendations,proposes a recommendation algorithm based on deep neural network.Firstly,establishes the corresponding relationship between project type and user project score for DNN training,and then establishes a multi classification deep neural network model with Dropout strategy.Finally,obtains the prediction score of target project by using model training and inverse coding.Experiments show that the new algorithm can provide reasonable recommendation based on the number of users'comments.
作者
程磊
高茂庭
CHENG Lei;GAO Mao-ting(College of Information Engineering,Shanghai Maritime University,Shanghai 201306)
出处
《现代计算机》
2018年第15期3-7,共5页
Modern Computer
基金
国家自然科学基金资助项目(No.61202022)
上海海事大学研究生创新基金资助项目(No.2017ycx061)
作者简介
程磊(1994-),男,硕士研究生,CCF学生会员,研究方向为数据挖掘、推荐算法。
