The Design And Implementation Of Point Of Interest Recommendation System Based On User Check-in Behavior Mining

The rapid development of wireless communication networks enables users to publish their geographic location in the form of sign-in,which is the point-of-interest(POI)in the user's sign-in data.This huge user check-in record constitutes the current Location-Based Social Network(LBSN).How to fully tap the potential value of user sign-in data has become a research hotspot in academia and industry.Some companies use the form of user sign-in to analyze the user's sign-in behavior data and mine the user's interest preferences,thereby recommending points of interest that they have visited but may be of interest to users,such as restaurants,scenic spots,and shopping malls.This form of task is called point-of-interest recommendation.Unlike traditional recommendation scenarios,there are implicit feedback and data sparsity problems in point-of-interest recommendation,and there are a lot of complex contextual information in point-of-interest recommendation scenarios,such as geographic location,time series,topic categories,Social relations,and these information are closely related to the user's sign-in behavior.Therefore,how to dig deeper into the relationship between user sign-in behavior and context information to improve the performance of the recommendation model is a problem worthy of in-depth study.The main research content of this article mainly includes the following parts:(1)This paper proposes a point-of-interest recommendation algorithm based on the geographic location and category information of points of interest.The algorithm model uses the category similarity between points of interest and the distance in geographic space to model influence and characterize the user For the preference relationship of different categories of points of interest,the influence of the categories of points of interest on the user's sign-in decision-making process is deeply explored.Experiments on two real data sets show that the recommendation performance of this algorithm model is better than other recommendation algorithms that incorporate contextual information.(2)This paper proposes a point-of-interest recommendation algorithm based on user sign-in area segmentation.The algorithm is based on the geographic location,popularity,and social relationship of the points of interest.Based on the weighted matrix decomposition model,the algorithm analyzes one of these factors.It also analyzes the regional characteristics of some users' sign-in,divides the user's sign-in area,and regularizes the user's social factors into the objective loss function.Through experiments on real data sets,it is shown that the algorithm proposed in this paper has a better recommendation effect than other recommendation algorithms.(3)This paper proposes a point-of-interest recommendation algorithm model based on the user's sign-in area and time neighbor modeling.The model uses a clustering algorithm that divides the user's area center to model the user's sign-in area and analyze the division The effect of the check-in area on the user's check-in surrounding points of interest,and analyzes the transfer characteristics of the user's point of interest preference within a certain period of time,and captures the dynamic changes of the user's preference overtime.Finally,experiments on two real data sets show that the algorithm proposed in this paper has higher recommendation accuracy than other recommendation algorithms.(4)For some users with unobvious user feature preferences,this paper proposes a point-of-interest recommendation algorithm that integrates multiple contextual factors,which can comprehensively analyze users'sign-in feature preferences from multiple angles,and alleviate some users'preferences due to scarcity of data.Describe problems that are not obvious.(5)On the basis of the four proposed algorithms for point-of-interest recommendation,this paper designs a point-of-interest recommendation system based on the characteristics of the four algorithms and the user's check-in behavior pattern,which can analyze the user's check-in behavior pattern.Make targeted recommendations for users with different check-in modes,improve the accuracy of point-of-interest recommendations,meet the needs of various users,and improve user experience.