融合知识图谱和图注意力网络的旅游推荐算法
Tourism recommendation algorithm based on knowledge graph and graph attention network
徐春 1王萌萌 1孙彬1
作者信息
- 1. 新疆财经大学信息管理学院,新疆乌鲁木齐 830012
- 折叠
摘要
为缓解旅游推荐模型面临的数据稀疏和冷启动的问题,提出一种融合知识图谱和图注意力网络的旅游推荐算法KRGAT(knowledge ripple graph attention network).借助水波网络从用户的历 史旅游行为和知识图谱中挖掘用户偏好增强用户特征表示,针对当前旅游项目特征学习的方法难以提取节点深层特征的问题,利用图注意力网络聚合相关度更高的邻居节点信息,增强旅游项目特征表示.实验在自建立的旅游数据集上与5个基线方法进行对比,其结果表明,KRGAT的精确率(P)、召回率(R)和AUC值分别提升了 5.73％、4.42％和1.42％.
Abstract
To alleviate the problems of data sparsity and cold start faced by tourism recommendations model,a tourism recom-mendation algorithm combining knowledge graphs and graph attention networks was proposed.RippleNet was used to mine user preferences from user historical tourism behaviors and knowledge graphs to enhance user feature representation.In response to the problem of difficulty in extracting deep node features using current tourism project feature learning methods,graph attention networks were used to aggregate neighbor node information with higher correlation to enhance tourism projects feature represen-tation.The experiment was compared with five baseline methods on a self-established tourism dataset.The results show that the accuracy,recall,and AUC values of KRGAT are improved by 5.73％,4.42％,and 1.42％,respectively.
关键词
旅游推荐算法/图注意力网络/知识图谱/水波网络/注意力机制/大语言模型/知识表示学习Key words
travel recommendation system/graph attention network/knowledge graph/RippleNet/attention mechanism/large language model/knowledge representation learning引用本文复制引用
基金项目
国家自然科学基金(62266041)
新疆财经大学科研项目(2022XGC073)
出版年
2024
NETL
NSTL
维普
万方数据