关联文本信息的网络表示学习推荐算法

Network representation learning recommendation algorithm associated with text information

通过基于随机游走的网络表示学习算法得到节点的低维嵌入向量,进而将其应用于推荐系统是推荐领域很流行的研究方向。针对当前基于随机游走的网络表示学习算法仅着重考虑了网络结构特性而忽略文本信息的问题,提出一种关联文本信息的网络表示学习推荐算法。首先在随机游走阶段,考虑到了节点文本间的相似度,联合结构和文本信息对下一游走节点进行筛选;然后在网络表示学习部分融合文本信息,引入注意力矩阵,对文本信息矩阵中的向量进行加权表示;最后将生成的节点向量应用于推荐系统。在实验部分,将所提算法与常见的3种算法在两个数据集上进行对比分析,并对所提算法进行了参数敏感性分析。实验结果表明所提算法在AUC评价指标上的性能优于另外3种算法,可见该算法在个性化推荐中的有效性。

In the field of recommendation,it is popular to get embedded vector of nodes with low dimension through the network representation learning based on random walk and then apply it to the recommendation system. Aiming at the problem that the current network representation learning algorithms based on random walk only consider the characteristics of network structure and ignore the text information,a network representation learning recommendation algorithm associated with text information was proposed. Firstly,in the phase of random walk,the similarity of text information among nodes was taken into account,and structure and text information were combined to filter the next walk node. Secondly,in the phase of representation learning,text information was fused,and the vector in the text information matrix was weighted represented by introducing the attention matrix. Finally,the generated node vector was applied to the recommender system. In the experimental part,the proposed algorithm was compared with three common algorithms on two data sets,and the parameter sensitivity of the proposed algorithm was analyzed. The experimental results show that the performance of the proposed algorithm is better than the other three algorithms in AUC(Area Under ROC(Receiver Operating Characteristic)Curve),which shows the effectiveness of the proposed algorithm in personalized recommendation.