Font Size: a A A

The Mas, Applied To The Study Of Distributed Data Mining

Posted on:2005-11-18Degree:MasterType:Thesis
Country:ChinaCandidate:A YiFull Text:PDF
GTID:2208360155954808Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Tranditional DM (Data Mining), basically, is a data-analyzing tool for local data at present and only can produce few of generalized or understood knowledge from local datasets. With the accumulating and distributing of data store, better methods or tools are needed greatly. Research of MAS (Multi-Agent System) has become a hotspot in the domain of DAI (Distributed Artificial Intelligence), and propels the progress of many research areas since 1990s. The author has taken part in Spatial Data Mining Research of CIS based on Web (PN:00JJY2059), which is imbursed by Hunan Fund of Natural Science since 2002, and presided over applying MAS to distributed data mining.This research aims at applying MAS to distributed data mining and brings forward correlative applied strategies. At first, Paper concludes the technologies of distributed data mining & Multi-Agent System, and studies framework and realization of JAFMAS (Java-based Framework for Multi-Agent System) detailedly. Then, paper designs and realizes a data-mining prototype system based on multi-agent framework. The prototype system extends JAFMAS, and mines distributed datasets successfully. Finally, paper brings forward conflict-negotiation algorithm for conflicts during the combination of global knowledge.This paper includes three parts. The first part discusses the basic theory of data -mining. In this part, the status of research, advanced technologies and challenges are concluded at first. Then, the paper emphasizes the theory and method of Decision Tree algorithm and realizes C4. 5 algorithm. At end of this part, paper introduces distributed data mining (DDM). At the second part of this paper, Author analysises a Java-Based Framework for Multi-Agent System (JAFMAS), and builds a prototype system for distributed data mining based on JAFMAS. The theories and frameworks of Multi-Agent System (MAS) are also generalized in this part. According to these theories, Paper brings forward conflict-negotiation algorithm for conflicts during the combination of global knowledge. Experiments are arranged at the end of the paper. The result of experiments shows that the prototype system...
Keywords/Search Tags:distributed data mining, Multi-Agent, distribute, C4.5
PDF Full Text Request
Related items