Design And Implementation Of Point Of Interest Recommendation System Based On Social Relationship And Geographical Location

With the continuous development of mobile positioning technology and the increasing development of social networks,more and more social software is integrated with geographic locations.Among them.there is huge space for technological development and commercial development prospects.Various algorithms on the recommendation of points of interest have become a hot topic in the recommendation field.At the same time,due to the processing of massive data,various big data technologies and recommendation architectures have also been fully practiced in system engineering.How to improve the accuracy of the recommendation algorithm and design a reasonable.efficient,and stable recommendation architecture has become an excellent subject that combines intelligent algorithms and engineering practices and also has strong practical significance and requirements.The following is the main work of this article:(1)Combined with the user's social relationship information and geographic location information,comprehensively analyzed the multiple similarities between users and the behavior intensity of the user's active area,and adopted the traditional collaborative filtering algorithm model and geographic location in the recommendation field.Based on the information features,a scheme combining offline training models and real-time feature data is designed.The interest point recommendation algorithm model of this paper is designed,and algorithm experiments are performed on real data sets,which proves the effectiveness of the algorithm.(2)A point-of-interest recommendation architecture with a data pre-processing module,an offline model training module,a real-time stream calculation module,and a user interaction module is designed.It can pull requests and logs from the server and perform data cleaning.Stored in a distributed storage system and model training on a computing platform,real-time recommendation requests for users located in a certain location,combined with offline model training to obtain an online model,and after the recommendation list is obtained based on the model,it is displayed to the user at the front end.(3)The system deployment on the local physical machine was realized,and the system was tested for functionality and performance,which proved that the system's functionality was well implemented and the performance was stable.This paper uses a software engineering method to describe the overall system in terms of requirements analysis,outline design,detailed design,implementation,and testing.The illustrations are rich and the tables are complete.Finally,this paper puts forward the future improvement direction of the system.