Knowledge Visualization Using Optimized General Logic Diagrams
论文作者：留学生论文论文属性：thesis
关键词：Knowledge Visualization Optimized General Logic Diagrams

 
   Abstract.      
   Knowledge Visualizer (KV) uses a General Logic Diagram (GLD) todisplay examples and/or various forms of knowledge learned from them in a planarmodel of a multi-dimensional discrete space. Knowledge can be in di®erent forms,for example, decision rules, decision trees, logical expressions, clusters, classi¯ers,and neural nets with discrete input variables. KV is implemented as a module ofthe inductive database system VINLEN, which integrates a conventional databasesystem with a range of inductive inference and data mining capabilities. This pa-per describes brie°y the KV module and then focuses on the problem of arrangingattributes that span the diagram in a way that leads to the most readable rule vi-sualization in the diagram. This problem has been solved by applying a simulatedannealing.      Keywords: knowledge visualization, GLD diagrams, diagram optimization, ma-chine learning.       
   Introduction       
   knowledge visualization tools are components of many commercialdata mining tools, such as Microsoft SQL Server 2000 Analysis Services,IBM DB2 Intelligent Miner, and Oracle Data Miner. They are also presentin noncommercial software, e.g. Weka [11] and YALE [10]. The authors arenot aware, however, of any tool with capabilities of Knowledge Visualizer(KV) described in this paper. KV is a unique knowledge visualization systemthat uses a General Logic Diagram (GLD) for representing examples andknowledge derived from them.         
   GLD, introduced by Michalski [6], is a planar model of a multidimensionalspace spanned over discrete variables (attributes). Each variable partitionsthe diagram into a set of disjoint areas corresponding to the values of thevariable. The lines separating these areas for a single attribute are calledaxes of this attribute. An intersection of the areas corresponding to singlevalues of each variable constitutes a cell of the diagram. Every cell of thediagram thus represents a unique combination of attribute values.