| The application and development of the Internet not only promotes the emergence and development of the new industries,but also provides opportunities and challenges for the traditional manufacturing industry.Conventional marketing methods,has been far from meeting the demand.With the rapid growth of Internet users and the rapid rise of the mobile Internet,the amount of Internet marketing data is growing,and the data structure is complex and irregular.Therefore,this thesis is based on the advantages of massive unstructured data storage and analysis of Hadoop cluster,and the flexibility to deal with different types of large data,put forward the framework of big data marketing system based on Hadoop.To realize the storage and processing of massive data,and combined with the massive data distributed crawler technology,model of crowd portraits of Internet users,the Internet user behavior data model and the analysis of the user’s interests and preferences based on a better guide to Internet marketing strategy and the implementation of the guidance.The main contributions are as follows:(1)Research background and demand;(2)Research on the technology of big data correlation;(3)The design of Internet data marketing system based on Hadoop;(4)The realization of Internet data marketing system based on Hadoop;(5)System deployment and Application.At present,the system has been completed and put into operation,the use of 16 servers,daily processing of 200 million logs.The system has been completed and put into operation,access to Internet user attributes and user habits have a more accurate understanding of the advertising and marketing decisions to provide a good support.In the manual: reducing the operator workload,improve work efficiency;in the marketing effect: due to the basis of user attributes of advertising methods,enhance the conversion efficiency and input-output ratio of advertising. |
