Research On Point Of Interest Recommendation Algorithm Based On Geographical Region And User Context Information

The progress of 5G communication technology has promoted the development of mobile Internet,and the Internet application platform has accumulated a large amount of user data during the development of the Internet.Among them,location-based social networking platform provides users with services such as punching in and sharing favorite places under the development of mobile Internet and GPS positioning technology,and these services have been integrated into people’s lives.In the process of using these social platforms,users generate a large number of check-in data.By mining these data,users’ preferences can be analyzed,so that users can predict their possible places of interest.Interest point recommendation system is a sub-problem of recommendation system research,which has high research value.Compared with recommendation systems in other fields,the data of interest point recommendation system is more sparse.At the same time,users’ travel decisions are influenced by various contextual factors such as geographical factors and time factors.Therefore,it is a great challenge to improve the accuracy of interest point recommendation.Existing interest point recommendation algorithms often can’t effectively combine geographical factors and social relationship factors.In order to solve this problem,this paper proposes a recommendation algorithm based on geographical regions.Geographical regionality is reflected in two aspects: first,the user’s clock-in data is regional in the latitude and longitude coordinate system,and different users have their own specific activity space.Secondly,due to the influence of urban construction planning,the distribution of interest points is regional.In order to make better use of geographical regions,based on the idea of collaborative filtering,this paper establishes similar user groups based on check-in data,and puts forward a calculation method of weighted virtual users.Then the mixed Gaussian method is used to cluster similar users.Finally,through the methods of similar user clustering and tensor decomposition,the geographical and regional information is fully mined and the places of interest are recommended for users.The traditional interest point recommendation algorithm often doesn’t consider the user’s current location and time,which leads to the generated recommendation content not conforming to the user’s current usage scenario.In order to solve this problem,this paper proposes a novel next interest point recommendation algorithm based on the user’s specific scene information.The algorithm makes full use of context information while mining the relationship between interest points,and recommends the next interest point that users need more.In order to better analyze the behavior pattern of users’ continuous check-in places,this paper fully explores the relationship between interest points by using the method of graph embedding.Then the graph embedding method is used to calculate the user’s interest characteristics with time and space factors.Finally,based on the user’s current situation,the user is recommended to be more interested in a certain time range.In this paper,experiments are carried out on several check-in data sets,and the experimental results show that the algorithm proposed in this paper has good performance.