摘要
提出一种基于位置的社交网络(LBSN)中利用深度学习的POI推荐方法。设计LBSN异构图,即UP2Vec模型,整合地理签到信息、用户社会关系和时间信息等语境。提出偏好增强谱聚类(PSC)算法,通过分析用户的各种语境信息获得多个维度的数据空间特征,使用谱聚类划分用户群体。利用谱嵌入增强的神经网络深度挖掘用户与POI之间的非线性关联,实现POI的高质量推荐。实验结果表明,所提方法性能优于对比方法,推荐准确率超过90%。
A POI recommendation method based on deep learning in location-based social network(LBSN)was proposed.The LBSN heterogeneous graph,namely UP2 Vec model,was designed to integrate geographic check-in information,user social relations and time information.A preference enhanced spectral clustering(PSC)algorithm was proposed,which obtained multidimensional data space features by analyzing users’context information,and spectral clustering was used to divide user groups.The nonlinear association between users and POI was deeply mined using the neural network enhanced by spectral embedding to achieve high-quality recommendation of POI.Experimental results show that the performance of the proposed method is better than that of the comparison methods,and the recommendation accuracy is more than 90%.
作者
刘旸
吴安波
李慧斌
LIU Yang;WU An-bo;LI Hui-bin(School of Management,Xi’an University of Science and Technology,Xi’an 710054,China;School of Mathematics and Statistics,Xi’an Jiaotong University,Xi’an 710048,China)
出处
《计算机工程与设计》
北大核心
2022年第10期2926-2934,共9页
Computer Engineering and Design
基金
国家自然科学基金项目(71673220)
陕西省教育厅专项科学研究计划——人文社科专项基金项目(20JK0232)
陕西省科学技术厅青年基金项目(2020JQ760)。
关键词
POI推荐
基于位置的社交网络
深度学习
偏好增强谱聚类算法
UP2Vec模型
谱嵌入增强的神经网络
偏好预测
POI recommendation
location based social network
deep learning
preference enhanced spectral clustering algorithm
UP2Vcc model
spectral embedding enhanced neural network
preference prediction
