Research And Application Of Railway Passenger User Portrait System Based On Big Data

With the rapid development of the railway and the rapid promotion of the “Internet+” business model,the Internet has penetrated into various service scenarios of railway passenger transport.The railway Internet users have experienced explosive growth,and a great deal of user behavior data has been accumulated.Constructing a railway passenger user portrait system through the user's behavior data can more accurately grasp the passenger group,delineate high-quality passengers,and formulate personalized and differentiated marketing and service strategies for different passenger groups.The rational allocation of railway passenger resources,to achieve the railway from the traditional marketing strategy to expand to the value of passengers based on the maximization of railway interests and passenger service quality optimization win-win strategy.This article first introduced the related social background and technology background,and deeply studied related technologies in the big data ecosystem,including distributed file system HDFS,distributed data warehouse tool Hive,and distributed computing framework Spark.At the same time,the overall architecture of the railway passenger user portrait system and the related algorithms for user tag calculation are studied and implemented.Finally,the application scene of railway passenger user portrait system was introduced and good application results were achieved.The main research results of this paper include:(1)Based on the actual requirements of the railway 12306 Internet ticketing system,station WIFI,advertising platforms,and Internet reservations,and combined with current technologies in the field of big data,the overall architecture of the railway passenger user portrait system was studied and designed.(2)In view of the characteristics of data generated by different information systems of railways,in-depth study and realization of structured data and unstructured data batch processing and real-time incremental processing methods.(3)Conducted in-depth research on railway passenger behavior data,designed a railway passenger tagging system,and conducted detailed studies on calculation methods for basic tags,statistical tags,and potential tags.(4)Based on the massive passenger behavior data,the MapReduce-based K-means algorithm,BP neural network algorithm and PassengerRank algorithm are studied and implemented,which improves the computational efficiency of potential labels(eg,user value).(5)According to the user portrait system of railway passengers,a system safety early warning system was designed to monitor the running status of the passenger user portrait system in real time so as to realize the rapid response of the operation and maintenance personnel to the abnormal operation of the system.(6)Based on the railway passenger user portrait system,several application scenarios were introduced,a complete set of platforms for data verification services for third parties was researched and designed,and good application results were obtained.