Design And Implementation Of Audit System In DSP Advertising Business

DSP (Demand-Side Platform) advertisement is a new Internet advertisement system. Every participant would often take part in exchanging their materiel, which made it very important to guarantee the quality of advertisement data. The implement of audit system provides the entrance of marking the properties of advertisement. It also provides the function of artificial audit and auto audit, which guarantee the quality of advertisement data and save human resource. Besides, system provides query interface in multiple conditions and multiple dimensionalities. Staff in the company can search all the advertisement data in time.There is an old audit system running online, which was designed for traditional advertisement business. It can only support limited amount of data. The new system should provide functions of DSP business and support more data. In this article, the author did the requirement analysis, including dividing the functions and analyzing the non-functional requirements. With that foundation, the preliminary design is finished by designing the architecture, the topology of the system and describing how to build the system. This part also contains the three level architecture of the system and the services depended on and the deployment online. What’s more, the detail design is also finished, which introduces the class design and process design of some important functions, the implement of methods in each function. Considering of the increase of data brought by DSP, some actions have been taken to optimize the performance in MVC architecture, and to defend the XSS(CSS attack, Cross Site Script) attack. To improve the flexibility and the maintainability of the system, the system uses the B/S architecture which is widely used in traditional business systems. J2EE is used to develop. The database is DDB, which is a distributed database system developed base on MySQL, and Velocity technic is applied to minimize the development in the view level. In order to speed up the query, the cursors, the caches and the sets are used a lot in the system. And Lucene is applied to deal with the big data querying. Until now, the whole audit system has been completed developing, which provides the reliable data for all the advertisement business systems. In this project, the author take parts in the whole process of requirement analyze, design, implement and publish. The work focus on design and implement of audit task distribute function, task query function, ad information query function, human audit function, auto audit function, audit statistics function and audit system management functions. The optimization and security parts are also finished by the author.Since doing DSP business, audit system has been developed eight versions with agile development. Until now all functions and methods of optimization in this paper have all published online. In these versions, more and more functions have been added and the performance is greatly improved. The last version has been running for two months steadily online. As for data, all DSP advertisement data has been marked category and risk type through audit system, which solves the problem of high imitation and fake advertisement.