A novel conceptual framework for mining high speed data streams

Many scenarios, such as network analysis, real time surveillance systems, sensor networks and financial applications generate massive streams of data. These streams consist of millions or billions of updates and must be processed to extract the useful information to enable timely strategic decisions. Mining data streams have many inherent challenges among which the most important challenges are adapting to available resources and assuring quality of the output result. Recent work in the area of data stream mining addresses these two challenges separately. Algorithms in the area of stream mining lack the combination of resource and quality awareness. That means, although they deal with resource adaptation, they do not take quality aspects into consideration. The purpose of this paper is to discuss the importance of resource adaptation and quality awareness with respect to data stream mining and then propose a novel framework that accounts for both quality awareness and resource adaptation. The proposed framework can be generalized for any stream mining technique.