Research On Scenic Spot Recommendation And Route Planning System Based On LBSN

The rapid development of society and the improvement of people's living standards have facilitated people's travel.Therefore,in recent years,China's tourism industry has achieved great development.At the same time,research topics on tourism have also received widespread attention.Tourism recommendations are one of many topics.The rapid development of the mobile Internet has provided data support for travel recommendation research.The data in the LBSN(Location-based Social Network)can not only reflect user preferences,but because of its spatial characteristics,the data in the LBSN can also reflect the rules of popular routes and user transfer patterns.Due to the rapid development of the Internet and its huge user population,the amount of data it generates is extremely large,which has caused most of the recommended methods to fail to achieve satisfactory results in terms of recall and accuracy.On the other hand,before traveling to a new city,route planning is basically a necessary task.Although there are currently many tourism websites that can provide data queries for tourists,this is undoubtedly increased due to the huge amount of data.The workload of the user,and cannot solve user preferences,constraints,etc.Especially when users plan travel itineraries,users need to spend a lot of time,but in most cases,users will still not get their own satisfactory results,because location selection and route planning is a difficult task for tourists,Personalized travel recommendation can not only reduce the workload of the user,but also effectively excavate the user's preferences and some hidden attractions,which can improve the user's travel experience.Based on the real data set in Foursquare,this paper makes related research from the two aspects of tourism recommendation and recommendation in tourism research.The main contributions are as follows:(1)Inspired by the TF-IDF method in text mining,this paper will use this method to mine the user's category preferences and integrate it into the filtering method based on the check-in frequency matrix.Experiments show that it can effectively improve the recommendation effect.(2)Aiming at the problem that different places have different attractiveness at different time periods and users' sensitivity to geospatial distance is different,this paper also takes the space-time factor as the influencing factor of the place.On the other hand,this article integrates the level of user consumption into the personality of the place In the current recommendation system,among the many types of interest points,the user groups are different.Therefore,by collecting and analyzing the consumption level of the user's consumption location,the user's preference can be more effectively mined.This paper uses adaptive kernel density estimation This method models the user's consumption level.Finally,a multi-factor location recommendation system that integrates the user's consumption level is proposed.(3)In the current research on personalized travel routes,most researchers focus on the mining of location attributes,and rarely consider the user's transfer mode.This paper uses the Markov model to mine the user's transfer mode law and treats it as An important factor in the choice of location;at the same time,a calculation method for user's personalized stay time and transition time considering the delay index is constructed.Finally,the state expands the path tree to plan the routes that meet the constraints,and finally it is compared with several different path recommendation algorithms Compare.