COMPONENT-BASED FRAMEWORK FOR MOBILE DATA MINING WITH SUPPORT FOR REAL-TIME SENSORS
Taneli Rautio, Perttu Laurinen, Juha Röning · 2009
Abstract: The increasing use of various mobile devices has shown that there is a need for mobile data mining applica-tions. While many existing data mining frameworks can be modified to handle data streams generated in real time, they are usually too complex and inflexible to be used in mobile devices. This paper presents Mobile Smart Archive, a component-based framework for data stream mining in mobile devices. The framework takes care of generic data mining operations, allowing the application developer to concentrate on implement-ing only application-specific functionalities. This reduces implementation time and generates fewer errors, since the underlying framework of the application is tested and robust. The presented framework is written in C++ and it extends the existing Smart Archive framework with support for mobile systems and real-time sen-sors. The benefits of framework-based applications in the mobile world are presented by building and testing a demonstration program in different computer architectures. In this paper we show that the MSA framework is suitable for building data stream mining applications for the hardware-oriented mobile environment. 1