Research On POI Recommendation Based On Interactive Behavior In Mobile Edge Scenarios

With the explosive growth of Internet information and the maturity of wireless communication technology,some location-based social networks(LBSNs)appear and flourish.Users check in at points of interest(POI)through mobile devices to share real world life experience on social networks.POI recommendation is one of the important tasks in LBSNs.Recommending unexplored POI to users plays an important role in people’s daily life.Therefore,the recommendation system based on POI has important research significance.The sparsity of user interaction behavior data has always been a very troublesome problem in the recommendation system,especially for POI recommendation.POI is a real place in the real world,and there is a complex interaction relationship between users and POI.Therefore,different from the traditional recommendation system,POI recommendation not only needs to consider user evaluation,but also needs to consider the interaction behavior relationship between users and POI,such as geography,time,social and context.At the same time,social data with user location information is highly sensitive,and the collection of data and the presentation of recommendations seriously depend on the quality of the network.When the network bandwidth is limited and unstable,the request and response delay of recommendation service is always unsatisfactory.Therefore,how to ensure the quality of POI recommendation service while ensuring user privacy and data transmission efficiency is the main research work of this thesis.The main contributions of this thesis are as follows:Firstly,this thesis proposes a POI recommendation framework based on interactive behavior in mobile edge scenario.The framework is composed of three parts: user,edge node and cloud POI recommendation system.Aiming at the security and efficiency of user and POI recommendation data transmission,a lightweight privacy protection recommendation method in mobile edge scenario is proposed;Aiming at the modeling problem of complex interactive behavior relationship between users and POI in the process of POI recommendation,a POI recommendation method based on knowledge graph in mobile edge scenario is proposed.Secondly,in order to meet the security and low latency requirements of POI recommended data transmission in mobile scenario,this thesis designs a lightweight data integrity verification system framework in mobile edge computing scenario.By combining algebraic signature strategy and optimization strategy based on matrix index,make the transmission and presentation of POI recommended data achieve a balance in terms of security and computing overhead,so as to realize the lightweight requirements.Thirdly,after the secure transmission of recommendation data,this thesis proposes a POI recommendation method based on knowledge graph in mobile edge scenario.Firstly,this method constructs a knowledge graph network containing the interactive behavior relationship between users and POI.On this basis,sub-graph networks with different factors are constructed.By using the local space of individual user preference and the global space of all user preferences,the relationship between the entities of each sub-graph is comprehensively reasoned to explore new POI.At the same time,the mask self annotation mechanism of random walk is used to selectively pay attention to the relevant high-order neighbors,Finally,the next POI recommendation is completed by fusing local and global space.Finally,based on the above methods,this thesis constructs a POI recommendation prototype system based on interactive behavior in mobile edge scenario.Its construction includes background introduction,development environment,requirements design,system architecture and overall process.At the same time,this thesis combines the prototype system with practical application,and verifies the effectiveness of the POI recommendation method proposed in this thesis through application demonstration.