用户-兴趣点耦合关系的兴趣点推荐方法

A POI recommendation approach based on user-POI coupling relationships

在基于位置的社交网络(LBSNs)中,如何利用用户和兴趣点的属性(或特征)之间的耦合关系,为用户做出准确的兴趣点推荐是当前的研究热点。现有的矩阵分解推荐方法利用用户对兴趣点的评分进行推荐,但评级矩阵通常非常稀疏,并且没有考虑用户和兴趣点在各自属性方面的耦合关系。本文提出了一种基于深度神经网络的兴趣点推荐框架,首先采用K-means算法对兴趣点按地理位置进行聚类,使位置相近的兴趣点聚为一类;然后,构建一个卷积神经网络模型,用来学习用户和兴趣点在各自属性(如用户年龄与兴趣点位置之间)上的显式关联关系;同时,构建另外一个神经网络模型,模拟机器学习中的矩阵分解方法,根据用户的签到行为,深入挖掘用户与兴趣点之间的隐式关联关系。最后,将用户与兴趣点之间的显式和隐式关联关系进行集成,综合表征用户-兴趣点之间的耦合关系,然后将学习到的用户-兴趣点耦合关系输入到一个全连接网络中进行兴趣点推荐。本文所提出的模型在Yelp数据集上进行了评估,实验结果表明该模型在兴趣点推荐方面有较高的推荐准确性。

In location-based social networks,a hot research topic is the development of methods for making accurate recommendations to users by taking advantage of the coupling relationships between user attributes(or features)and points of interest(POIs).Current POI recommendation approaches mainly leverage the matrix factorization technique based on user ratings of POIs.These approaches routinely confront the problem of rating matrix sparsity,and do not consider the coupling relationships between the respective attributes,such as the descriptive information and comments of users and POIs.In this paper,we propose a POI recommendation framework based on a deep neural network.First,a K-means algorithm is used to cluster POIs by their geographical locations,so that POIs with high location proximity are grouped into one category.Then,a convolutional neural network model is designed to identify the explicit correlations between the respective attributes of users and POIs,for example,the correlation between the age of users and the location of POIs.Another neural network model is designed to simulate the matrix factorization method in machine learning.This model can examine the implicit relationships between users and POIs based on the users’POI check-in data.Lastly,the explicit and implicit relationships between users and POIs are integrated to comprehensively represent the user-POI coupling relationships.The learned user-POI coupling relationships are input into a fully connected network for recommendation.The proposed model was evaluated on the Yelp data set,and the experimental results show that it achieves high recommendation accuracy.