A Data Mining Platform Of Public Opinion

With the development of computer and network technology, Internet becomes one of the most important media of information spreading. Since this medium has the features of direct, sudden, bias, loose and so on, it needs to be supervised and controlled by the related departments by the efficient public opinion analysis system. Traditional public opinion data mining system, however, not only has the shortcomings of inefficiency, low-real-time, long period and high cost of construction and poor scalability, but also has a poor performance in maintenance. Current system is no longer able to meet the demand of the strong real-time data, high dimension and uncertain semantic features.In order to make a breakthrough in the system, this paper is engaged in making a combination of cloud computing platform and data mining platform. It removes the constraints of computing power by using the theoretically unlimited scalability of cloud computing platform. Also, it is designed into a service-oriented model by using SOA architecture, which is introduced to manage the large number of cloud-based operations and related calls, and solve a lot of confusions and high application management problems. Obviously, changes in system requirements become fast, efficient and low-cost through the SOA architecture. This system is designed based on the service registry and the enterprise service bus pattern, and the implement of SOA based on Web services enhances the characteristics of virtualization and service-oriented of the system.This paper compares the existing framework of distributed data mining platforms, before proposing a design of cloud-based public opinion data mining platform referring their respective advantages. The system is designed into serial layers, such as public opinion data storage layer, the mining task allocation layer, application layer and SOA applications. In addition, algorithm implantation system is designed to make the system more flexible, for users can upload self-designed algorithms to the system platform according to their needs and system permissions. The plug-up inquiries docking system and its interfaces designed in this paper make the system more dynamic management and maintenance.Finally, based on the discussion of the system above, the last section presents the user platform of public opinion data mining platform and the interface of mining algorithms. Experimental data and the user experience shows that our data mining platform of public opinion based on cloud computing is a greatly improved platform than traditional ones.