Research And Implementation Of Audio Advertising System Based On Spark

At the moment when everything is connected,audio advertisements can be embedded in many media,such as building elevators,outdoor broadcasting,etc.Audio advertisements can be embedded into people’s lives around the clock,from before and after the weather forecast is played by the intelligent robot at home in the early morning,to the soothing music played in the car during the commute,to listening to novels on the mobile phone in spare time,etc.With the help of the unique attributes of the sound medium,diversified scene dissemination,we can maximize the exposure of brand information.At present,audio advertising has a huge market blank and broad development prospects.This thesis investigates the business process and technology of advertising delivery,summarizes the characteristics of the audio advertising delivery system,such as different usage scenarios,different heterogeneous network environments,time requirements for advertising playback,etc.,and completes the system requirement analysis.On the basis of requirement analysis,the system design is completed.The system is divided into two subsystems: audio advertisement management and delivery,and data management.Audio advertisement management and delivery includes registration and login of audio advertisers,delivery management,audio material management,delivery engine,etc.The main functions of data management include log collection,data report,etc.This thesis analyzes and studies the key issues in system development,such as voiceprint recognition technology,targeted advertising technology,and anti-fraud function.The technologies used in this thesis mainly include Kafka distributed message queue,Flume distributed log collection tool,HDFS distributed file storage,Spark computing framework,etc.It achieved the main functions of this system and tested it.The realization of the system opens the way of auditory advertising for advertisers and provides a new solution.