基于动态知识图谱和深度神经网络的会话推荐方法

Session recommendation method based on dynamic knowledge graph and deep neural networks

基于深度神经网络对会话进行建模的方法在处理会话序列数据时忽视了项目动态知识和会话中的操作行为,影响了推荐的准确性问题,为此提出基于动态知识图谱和深度神经网络的会话推荐方法。分别采用图神经网络和循环神经网络学习项目序列和操作序列的特征表示,结合动态知识图谱的项目知识进行建模以达到动态推荐的目的。实验结果表明,该方法能够提高推荐结果的准确性,更为有效预测用户的下一个交互项目。

To solve the problem that the method of session modeling based on deep neural network ignores the project dynamic knowledge and operation behavior in the session when processing session sequence data,which affected the accuracy of recommendation,a session recommendation method based on dynamic knowledge graph and deep neural network was proposed,graph neural network and recurrent neural network were used to learn the feature representation of item sequence and operation sequence,and the item knowledge of dynamic knowledge graph was combined,modeling was carried out to achieve the purpose of dynamic recommendation.Experimental results show that the proposed method can improve the accuracy of the recommendation results,and predict user’s next interactive item more effectively.