Design And Implementation Of Restaurant Recommendation System Based On User Check-in Correlation Information

In recent years,the rapid development of mobile Internet promoted the location based social network(Location-Based Social Networks,LBSNs),with the rapid development of restaurant recommendations services with the booming development,but a growing number of restaurants catering to join the service platform,has brought the information overload problem,user is facing huge amounts of restaurant information failed to quickly locate interested in his restaurant.As a means to solve the problem of information overload,restaurant recommendation system is used to select the restaurants that users may be interested in from a large number of restaurants.This paper has conducted relevant research on restaurant recommendation,and the specific contents are as follows:(1)Considering that users' dietary preference is not invariable and users show their preference changes in different cities or different geographical regions,based on users' preference changes in different geographical regions,a restaurant recommendation algorithm AUPDM adapted to users' preference migration is proposed.Based on the context information in the user check-in record,the preference characteristics of users in different geographical regions are mined,and the role preference of users in geographical regions is modeled.Through the comparison experiment on the public data set Yelp,the results show that AUPDM has certain improvement in the recommendation accuracy rate and recall rate.(2)Through the effect of user check-in record mining time on the user's choice of restaurant,the time not only affects the geographical area of the user's choice of restaurant,but also affects the theme of the user's choice of restaurant.Mining user history checkins associated information in time characteristics,this paper proposes a fusion time characteristics TAR/R restaurant recommendation algorithm,modeling the user preference model about the time,using a model which generates the probability integration time characteristics,and finally on the public data sets,Foursquare contrast experiment,the results show that TARR can make better effect on recommendation,recommend the precision rate and recall rate has certain ascend.(3)To analyze user activity of geographical area,mining user preference characteristics about dinner on geographical area,and then analyzed the influence of the time for the user to select a restaurant location,and then through the fusion area and time probability model modeling user personalization preferences,put forward a kind of based on geographic region and the user check-in time GTRR restaurant recommendation algorithm,finally through contrast experiment on Yelp public data sets,the results show that GTRR can effectively improve the precision rate and recall rate.(4)Combined with several restaurant recommendation algorithms proposed,a restaurant recommendation system based on user check-in correlation information is designed and implemented.The recommendation module of the recommendation system integrates the restaurant recommendation algorithm proposed in the previous chapter and provides restaurant recommendation services to users through the user interaction module.