A Single Pass Algorithm of Finding Frequent Vibrated Items over Online Data Streams

• Main Authors: Chiao-Tzu Chen, 陳巧慈
• Other Authors: Guanling Lee
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/35171707503654796011

碩士 === 國立東華大學 === 資訊工程學系 === 94 === Data streams are data items which generate unbounded and continuously, and data in the streams change all the time. In order to detect the vibration of data item’s quantity, we propose a single pass algorithm MVI (Mining Vibrated Item) for mining vibrated items over online two data streams. The change of the data item can be reported at once by measuring its vibrated slope. Not only the change of the data item will be found, we also trace the period in which the data item is frequent vibrated. The frequent vibrated item and its corresponding period is reported by our algorithm MFVI (Mining Frequent Vibrated Item). The theoretic analyses show that the frequent vibrated periods outputted by our algorithm have the guarantee that the ratio of the number of times the items regarded as vibrated items to the length of the period is no less than the threshold. And our experimental studies on synthetic data show that our algorithms detect the vibrated items accurately and efficiently.