Research On Key-techniques For Multiple Data-Streams Processing

Data steam is regarded as hotspots in the research area of data mining recently. Along with the rapid development of information technology, the demand for stream data processing is proliferating rapidly. Data stream is featured with mutability, infiniteness and succession. In the multiple data streams environment, the coincidence relations between data streams are significant. In this dissertation, a series of key-techniques for multi-streams processing are researched such as the coincidence problem and storage problem and forecast problem etc. The main contributions of these works included as follows:(1) Introduces concept of total ordering to describe the characteristic of data streams, and proving that data stream satisfies total ordering relation on time dimension.(2) Proposes new algorithms to mine synchronous or asynchronous coincidence pattern in multiple data streams. The main contributions of this work are as follows: Formally describes a series of conception such as synchronous coincidence and asynchronous coincidence. Investigates the filter technique of Haar Wavelet and applies it to mining synchronous or asynchronous coincidence pattern in multi-streams. Applies the Wavelets coefficient series to the measurement of synchronous coincidence and asynchronous coincidence between data streams. Proves a series of theorems to ensure the validity of measure synchronous or asynchronous coincidence such as Local center distance theorem, Coincidence equivalence theorem, Distance equivalence theorem, Strong coincidence determinant...