Research On The Tourist Attractions And Route Recommendation Based On Crowd Geographic Data

With the rapid development of internet technology and the improvement of per capita consumption level.traveling outside has become a fashionable consumption activity for people to relax.But how to timely and accurately choose the right tourist attractions and routes to save time,money and get a better travel experience is the current problem.Currently,the domestic mainstream travel websites provide popular and universal recommendations in the attractions and route recommendations.In order to make up for the recommendation results status of no targeted,this article is based on crowd geographic data as a data source and propose an attractions recommendation algorithm which is based on the attractions clustering and user score and a route recommendation algorithm which is based on multi-dimensional feature clustering.This article use ArcGIS for android technology,database technology and front-end development technology as technology support to design and develop a travel recommendation system.Acquisition and processing of crowd geographic data.This article takes Qingdao as the research area and mafengwo travel website as data source website,and use the GooSeeker website data crawler tool to crawl attraction information,score,and travel text data and use the python and jieba text segmentation tool to preprocess the data.Finally,the processed scenic spots information,attractions score,travel routes,and quantified attractions tag and travel feature data are stored in the database.Attractions recommendation based on the attractions clustering and user score.Because of the traditional collaborative filtering recommendation algorithms has data sparse and cold start problems,this article firstly uses K-Means clustering to cluster spots by the attractions tag.At the same time,based on the original similarity of the score,propose the score reliability,which is designed by recommendation credibility and quality credibility,so as to improve the accuracy of the recommendation.Finally,linear combination of score similarity and score reliability as the final recommendation method and output of the results by TOP-N and use the MAE value to verify the recommendation algorithm.Experimental results show that this algorithm not only reduces the data sparse,but also improve the recommended precision and have better stability.Route recommendation algorithm based on the multi-dimensional feature clustering.Because of the travel website does not consider the user’s own situation in the route recommendation,this article uses the attribute weighted K-Means clustering to cluster the travel time,cost,number of days,people and whether or not the hotel and food in the source data.This make the users in the same category have similarity in the choice of travel routes.In order to improve the accuracy of route recommendation,this article uses Apriori algorithm to mine frequent attractions and then calculate the popularity of the recommended route.Finally,the recommended route is TOP-N output based on the popularity result and use the precision rate,recall rate and F-Measure value to verify the recommendation algorithm.Experimental results show that,this algorithm not only meets user needs,but also has higher accuracy.Implementation of tourist attractions and route recommendation system.The system realizes the recommendation and output of tourist attractions according to the user’s preferences and attractions score.At the same time,the system can recommend travel routes that meet the needs of users according to the feature of travel.The system use a list and map visualization to show the recommendation result.Through experimental analysis and system testing,applying the method proposed in this article can accurately and effectively carry out the recommendation of attractions and routes,which is of great significance in satisfying users’ own needs and personalized recommendations.