基于传播树的微博推荐模型

针对目前微博推荐模型未考虑传播特征的问题，提出一种基于传播树的微博推荐模型。首先利用树结构对微博传播特征进行表示，由内容、时间和用户三方面特征构成树的节点，以微博的转发和评论关系作为树的边；然后基于节点间关联关系和层次关系分别计算待评估微博传播树与目标用户每棵微博传播树的传播路径相似度和传播层相似度，以此量化两棵传播树间的结构相似度；最后根据相似度大小对所有待评估微博进行排序，生成推荐列表，实现微博推荐。实验结果表明，与未考虑传播特征的微博推荐模型相比，该模型在准确率、召回率和F1值上分别提升13.0%、9.6%和10.7%，合理利用微博传播特征可以提升推荐结果的可靠性，增强用户体验感。

Microblog recommendation model based on propagation trees

The paper proposed a new microblog recommendation model based on propagation trees to address the issue that the current microblog recommendation models haven''t considered the propagation feature. Firstly, it used the tree structure to represent the microblog propagation feature, with the content, time, and user features forming the nodes of the tree and the forward and comment relationships of the microblog as the edges of the tree. Then, it calculated the propagation path similarity and propagation layer similarity between the to-be-evaluated microblog propagation tree and each microblog propagation tree of the target user. It based on the association relation and hierarchical relation among nodes to quantify the structural similarity between the two propagation trees. Finally, it sorted all the to-be-evaluated microblogs according to their similarity and generated recommendation lists to achieve microblog recommendations. The experimental results show that the model improves by 13.0%, 9.6%, and 10.7% in precision, recall, and F1-value, respectively, compared with the microblog recommendation model without considering the propagation feature. The reasonable use of the microblog propagation feature can improve the reliability of recommendation results and enhance the user experience.