Font Size: a A A

The Research On Mining Methods About The Average Of Network Consensus

Posted on:2008-08-21Degree:MasterType:Thesis
Country:ChinaCandidate:X D ZhaoFull Text:PDF
GTID:2178360245498162Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Internet is the main carrier, with which people gain and pass information. Those abundant and different kinds of data form a very rich information source on network. With the high development of internet, the research of network consensus is facing new problem and chanllenge. Network consensus incarnates the concentrative and true view of the majority to network abnormity accidents. So with the help of getting information from internet itself, it is necessary to find different coming crisis factors and establish a perfect network consensus detection system.A network consensus detection is dimidiate into network abnormity detection and network trend detection. For the former detection, we use detecting technique of frequent items from data stream, and bring forward a novel digest structure with wavelet set and propose an improved burst detection algorithm. On this condition, necessary association analysis is taken out. As to the latter one, after using existing average system in correlative field for reference, we propose a brand-new concept named Average of Layered Network Consensus.Considering the complexity of clustering to Chinese character, firstly we disassemble the documents with key words into Chinese words, and then combine manual classification, statistic frequent item mining and clustering to get frequent items, which is defined as the key words of consensus.Then, contraposing the key words of consensus we draw frequency graphs, and do necessary burst analysis and association analysis. For the burst change, we use methods combined wavelet transform and gliding window technique, and broaden the area for data analyzing. For above-mentioned change, we improve the traditional burst detection algorithm, then propose the burst change detection algorithm. Besides that, we present the necessary performance estimates. Meanwhile, we make more association analysis with chosen key words of consensus, and give necessary support to establishing the average of consensus system.Finally, after using existing average system for reference and comparing with existing average of Nisac, we propose a definition of Average of Layered Network Consensus, and bring forward a theoretical structure of the average of network consensus original system.
Keywords/Search Tags:Network Consensus, Average of Consensus, Frequent Item Mining, Burst Detection, Wavelets Analysis
PDF Full Text Request
Related items