Design And Implementation Of Recommendation System Based On Users' Check-in Tradjectory

With the rapid development of mobile Internet,the rapid popularization of intelligent terminals and the continuous maturity of personalized recommendation technologies,location-based recommendation services have gradually been incorporated into people's daily life.At the same time,the space-time information that people share through various mobile smart devices is usually driven by the user's interest,so the user's space-time information is transformed into a trajectory mode,and the similarities between users are tapped from these trajectory modes to achieve more accurate Recommended service.This paper focus on the location information on the mobile terminal and studies the existing personalized recommendation techniques.Aiming at the advantages,disadvantages and usage scenarios of the current mainstream location service recommendation algorithms,the original point-of-stay extraction algorithm is improved and the semantics of the stay point in the user trajectory are more accurately obtained.At the same time,based on the traditional trajectory similarity,new matching conditions of semantic points are added and the algorithm of user similarity is improved.The evaluation of the performance of the algorithm through the evaluation criteria in information retrieval shows that the improved algorithm has been improved in the recommended accuracy.Then,this paper uses a recommender system to implement a recommender system that includes two algorithms based on demographic recommendation and recommendation based on user semantic trajectory similarity.According to different recommended scenarios,switch the recommended algorithm,effectively solve the user cold start problem.Finally,according to the functional requirements of the system,a complete functional module is designed and implemented.Paper work through real users to connect WiFi hotspot location record data set for experimental verification.The verification results show that the proposed system based on user's sign-in trajectory proposed in this paper has more excellent performance in terms of accuracy and stability.