Research On Location Recommendation Algorithm Based On Spatiotemporal Characteristics And Trust Relationship

With the popularization of intelligent mobile terminals and the rapid development of location based social networks,people will produce a large number of check-in content with geographic information every day,but also produce the problem of information overload.These multi-source heterogeneous data contain rich user behavior and personalized demand information,which makes place recommendation research widely concerned by scholars at home and abroad.As a direct channel connecting the online and offline world,user check-in behavior has helped to develop many personalized positioning information services.Therefore,accurate and personalized location recommendation is particularly important.However,compared with the product recommendation system,the places that users have checked in are relatively few,the data is more sparse and the scalability is limited.In this paper,the characteristic attributes of user check-in data are analyzed,and the location recommendation algorithm of the location social network is deeply discussed by using the convolutional neural network and the sorting learning model.Firstly,the location recommendation algorithm under the influence of time and space characteristics is explored.Firstly,the correlation between user behavior and time and space factors is studied by analyzing user check-in from time and space levels.Secondly,the convolutional neural network is used to construct the temporal and spatial representation of the location to capture the user's preference for the location and improve the accuracy of location prediction.Subsequently,we explore a location recommendation algorithm which integrates trust relationships and comment text.Firstly,the trust between users is calculated by combining the trust relationship of social networks.Secondly,the topic information of the comment text is incorporated into the similarity calculation to obtain the similarity of the comment text.Finally,a list-level sorting learning model combining trust and similarity is proposed to improve the quality of the recommendation list of places.This paper utilizes the convolutional neural network and the sorting learning model to mine user check-in and social relations to improve the quality of recommendations and effectively improve the user experience of location-based social networks.Location recommendation is not only helpful for local residents and tourists to explore the unknown interesting places in the city,bringing convenient and comfortable life experience;At the same time,it is helpful for merchants to find and attract potential customers and get the opportunity to increase revenue.