dl环境下的信息资源理及知识发现研究-管理科学与工程专业毕业论文.docx

哈尔滨工程大学博士学位论文 Abstract Aiming at information resources constructjoa of digital library， the methods of information organizing and data mining are studied． Fi rst．the characteristics of information resources of digital 1ibrary are giyen，the restrictions of general cata]ogue in describing networks i nformation are shown．The metadata methods of describing networks information resources are expatiated．The definition，type， structure，coding language of metadata are presented．The interaction of metadata is researched．The application of typical metadata Dubl in core is analyzed．The element formation of Dubl in COre iS studied．The differences between Dubl in core and general catalogue are compared． The networks information resource catalogue is discussed． SecondlY，for data mining aspect of networks information，fuzzy vector space model is shown．The methods of fuzzy characteri stic extrac Li ng and fuzzy characteristic vector constructi ng are presented． Three clustering algorithms are presented，these a1 gotithms are as follows：k-means clusteri ng algorithm．kernel c】ustering a】901，ithm and ant clusl：er intelligence clustering algorithm．These ciustering algorithms are based on fuzzy vector space model．Applying theory of fuzzy logic，neural networks and machine 1earning in art ificial i ntelligence fields，three classifi cation algotithms are presented， these algorithms are as follows：document automatic classification method based on weighting fuzzy reasoning networks，document automatic classification method based on self—organization feature mapping networks，document automatic classification method based on suppert VeCtor machine．Each of algorithms gives detailed classification principle and approach．The availability of the algorithms 1s proved by experiment． Finally，a few counterJI】easures of reinforcing DL management and knowledge discovery are presented． Keywords：Digital library：Resource management：Knowledge di scovery： Data mi ning：Algori thm design．  哈尔滨工程大学 学位论文原创性声明 本人