众筹项目推荐:面向隐式反馈的深度学习协同过滤

目前,在推荐系统研究中,用户的隐式反馈,以及极度稀疏的数据,已成为影响协同过滤推荐效果的主要问题.针对这一现象,本文提出了深度学习协同过滤算法,先利用卷积神经网络,对用户-项目矩阵的隐层特征进行学习,再结合协同过滤,对用户-项目的交互信息进行建模,并将两种特征融合预测推荐列表.以众筹平台的数据为实验对象,比较模型中各参数对推荐效果的影响,并设计与基线方法的对比实验.实验结果表明:均匀采集负反馈,并在一定卷积层数的网络中,数据稀疏度越高,效果越好;对比基线方法,本文提出的算法在公开数据集(Yahoo!Movie)上取得了最好的推荐结果.本文提出的算法有助于提高众筹平台的融资成功率,同时也丰富了推荐系统的研究体系.

Crowdfunding Project Recommendation:a Deep Learning Collaborative Filtering Approach for Implicit Feedback

Recently,in the research of recommendation systems,the users'implicit feedbacks and the extremely sparse data have become the primary problems influencing the recommendation performance of collaborative filtering.In reaction to this problem,this paper proposed a deep learning collaborative filtering algorithm.convolutional neural network was utilized to study the implicit features of user-project matrix at first,and then integrated with collaborative filtering to establish the model of user-project interactive information,and finally the projects recommending list for users was predicted based on such interactive information.The data of a crowdfunding platform was selected as the experimental object,and the comparative experiments were designed to analyze the influence of each parameter in the model on recommendation performance,and to compare the performance of the proposed method in this paper and the baselines.The experimental results showed that with the balanced collection method for implicit feedback,and a certain number of convolutional layers,the higher the data sparsity,the better the effect;compared with baselines,the proposed method in this paper achieved the best performance of recommendation on the public dataset Yahoo!Movie.Therefore,this research not only is of help to improve the financing success rate of crowdfunding platform,but also enriches the research of recommendation systems.