| In recent years,with the wide application of intelligent devices and the rapid development of mobile networks,users are used to sharing their own experiences with close friends,so a large number of user trajectory data with spatial and temporal characteristics are constantly generated.The large scale of the trajectory data contains a wealth of information,which brings a huge challenge to the recommendation.Nowadays,people urgently need to solve the problem of how to recommend trajectory data quickly and accurately.Clustering technology is an important method to obtain trajectory data information.By clustering trajectory data,we can mine and analyze the users' friend information,activity rules,behavior patterns and provide users with high-accuracy recommendations that meet their requirements.There are three parts to user interest area recommendation method based on trajectory data.First of all,an improved density peak clustering method is proposed for mass trajectory data.This method improves the problem that the initial parameter truncation distance and the time complexity of selecting the initial clustering center point are too high in the Clustering by Fast Search and Find of Density Peaks.The improved clustering algorithm is used to preprocess the data and the individual user interest area is obtained.Secondly,for the extraction of public interest areas,the geographical relationship and social relationship between interest areas are integrated to build a relationship model and according to the relationship matrix,the public interest areas are divided and extracted.Finally,for the region of interest recommendation,according to the density of social relations,the public areas of interest are arranged in descending order from high to low and the threshold is set,the public areas of interest higher than the threshold are collaborative filtered and recommended.At the same time,the feasibility and effectiveness of the above methods are verified by experiments.The experiment uses GeoLife data set,which is a real user history trajectory data set provided by Microsoft Asia Research Institute.Through experiments on the dataset,it is proved that the proposed clustering method is better than the Density-Based Spatial Clustering of Applications with Noise and Clustering by Fast Search and Find of Density Peaks in accuracy and is better than the above two clustering methods when considering both time and accuracy.In the part of recommendation,we compare this method with probability factor algorithm and singular value decomposition algorithm and the accuracy and recall rate are all higher than the two algorithms.It is proved that the proposed method can improve the accuracy of existing user interest area recommendation method based on trajectory data to a certain extent. |
