摘要
利用深度学习在特征提取方面的优势,挖掘嵌入用户和项目信息中的隐藏信息,改善传统协同过滤算法中存在的数据稀疏性及冷启动问题,将提取到的特征信息运用协同过滤算法评分预测,并且考虑用户的兴趣漂移情况和物品流行度情况,增加用户时间偏置和项目时间偏置,使算法具有实时性。最后与多种算法进行对比实验,通过计算RMSE,评估算法的可行性与有效性。实验结果表明,基于深度学习的协同过滤推荐算法可行有效,能缓解传统协同过滤算法中存在的数据稀疏性、冷启动问题,具有实时性,提高推荐准确率,具有良好的推荐效果。
The advantages of deep learning in feature extraction is utilized in mining hidden information embedded in user and item information.The data sparsity and cold start problems in traditional collaborative filtering algorithms is alleviated by using the extracted feature information in score predictions and considering the user’s interest drift situation and item popularity situation to increase the user time offset and project time offset,so that the algorithm can run in real time.Finally,a comparison experiment with various algorithms is carried out,and the feasibility and effectiveness of the algorithm are evaluated by calculating the RMSE.The experimental results show that the collaborative filtering recommendation algorithm based on deep learning is feasible and effective,and can alleviate the data sparsity and cold start problems in the traditional collaborative filtering algorithm with real-time performance,which improves the recommendation accuracy,and has a good recommendation effect.
作者
刘航
李锡祚
LIU Hang;LI Xizuo(College of Computer Science and Engineering,Dalian Minzu University,Dalian 116000,Liaoning,China)
出处
《智能计算机与应用》
2020年第8期100-104,共5页
Intelligent Computer and Applications
关键词
深度学习
协同过滤
数据稀疏性
冷启动
实时性
deep learning
collaborative filtering
data-sparsity
cold-start
real-time