| 摘要 |
A grid-based mass data dividing device and a method thereof are provided to reduce a data dividing time by generating/training clusters to divide mass data, and removing the clusters not representing the data or having lower weight from the trained clusters, and increase stability by allocating grid resources to a data dividing process dynamically. A data divider(10) includes a resource managing module(11) dividing the available grid resources, a threshold adjusting module(13) adjusting threshold depending on a result of the divided data while dividing the data after initializing the threshold, and a result testing module(15) transferring the result by checking the result. A data intermediary(30) includes a plurality of resource intermediating modules(31) transferring the grid resources and the threshold, and merging/transferring the divided data to the result testing module. A plurality of data dividers(50) train the clusters by receiving the grid resource/threshold and dividing the mass data to the clusters, and removes and returns the clusters having an abnormal value to the data intermediary.
|
