Research On Distributed Knowledge Discovery System Based Mobile Agent

With the fast growth of calculation and network , a number of distributed surroundings appear ,such as intranet, wireless network etc . It is a hotspot in the area of knowledge discovery that how to discover and pick up valuable knowledge from distributed-isomerously data.This thesis aims at the shortcomings on expansibility , validity, transplant quality, adaptability, extending quality in current knowledge-discovery system , so we proposed Distributed Knowledge Discovery System based Mobile Agent ~ MA-DKDS.It is advanced in these abilities such as calculational resource deployment, distributed and collateral calculation , asynchronous communication and network resource optimization . And this system can select, integrate , mine data from distributed-isomerously data source.In the first chapter, the research and applying actualities about knowledge discovery and knowledge-discovery system are given, and then the study objectives of this thesis are presented: the requirement of distributed knowledge-discovery system , the architecture of MA-DKDS, the communication mechanism between mobile agents, data predisposing and data mining.In the second chapter, based introducing traditional architectures of knowledge-discovery system and analyzing current popular mobile agent technology , this thesis presents Distributed Knowledge Discovery System based Mobile Agent ~ MA-DKDS and introduces its architecture and work mechanism.In the third chapter, we research communication mechanism of system . Firstly interaction-process model is given and selects the appropriate connection way between agents ; Secondly due to use KQML language to realizing communication in high level of system , this thesis introduces this language simply and use XML technology to expand KQML's performative ; lastly the structures of agent and agnet control center, interaction process control principles are given.In the fourth chapter , this thesis introduces objective and procedure of data-predisposing procedure during knowledge discovery , researchs data-integration ways and data difference during integration , presents data-integration plat based virtual database and then introduces the design idea about plat's architecture and structure of every part .Lastly provides the detailed solutions about how to deal with data-difference problem during integration process.In the fifth chapter, this thesis introduces the peculiarity, task and policy about distributed data mining . we describe basic concepts , theories and main arithmetic steps of synchronous arithmetic FDM and asynchronous arithmetic DMA . And then compare peculiarities and workoccasions about these two arithmetics . Lastly we select asynchronous arithmetic as our system's Data-mining arithmetic.In the sixth chapter , this thesis introduces the application instance of MA-DKDS prototype system which was named "the research and application in machine-accessory industry about balanced producing management and innovative designing product which supports management on knowledge property right ". In this project, there are three aspects which use MA-DKDS prototype system's function : classifying knowledge , forecasting market , Integrating data between systems .In the last chapter , this thesis generalizes the whole thesis and prospects the next work content.