User Behavior Analysis And Attraction Recommendation Algorithm Research In Location-based Services

With mobile devices becoming increasingly popular in our daily life,geo-tagged data are widely shared.By exploring this type of data,much progress has been made by scholars to provide better location-based services for users.Among these services,tourism recommendation is one of focused directions in recent years.Different from book or film recommendation that oriented by content,tourism recommendation has stronger correlation with temporal-spatial factors.It is a derivative area in recommendation system because of the mutative demand.In this paper,we aim to provide better attraction recommendation for users,and help people enhance their impression on cities with a dataset collected from Flickr,which is a photo sharing social media platform.Since user behavior is the important basis for recommendation,we analyze it in detail.First,with the help of Wave Cluster Algorithm,we extract attractions around the world from raw data according to two level iteration of cities and attractions,considering accuracy of GPS technology meanwhile.Next,to mine activity pattern of user group,as well as to discover the frequent tourist attractions and their relations,we build visit sets for each user and introduce Apriori into attraction relation mining area.Visualization on digital map is also provided for better exploration.Our evaluation experiment on cities about reverse address resolution,and comparison experiment on attractions and relations with reality show that our results are believable and highly consistent with the reality.Finally,we design and complete our experiments on tourist attraction recommendation from three aspects of user group visit pattern,individual preference,and integrated.Calculating several metrics,our evaluation experiment shows that recommendation based on user group visit pattern has better effectiveness and execution speed,while recommendation based on individual preference has better precision.Besides,our integrated recommendation scheme can learn the strengths of the former two.It is the best among the three kind of attraction recommendation schemes.