Multi-Users Online Visual Data Mining System

Data mining refers to mining or extracting knowledge from the massive data set. Its emphasis is placed on uncovering interesting data pattern hidden in the data. Because of data mining algorithm itself complexity, it is difficult to understand and control these algorithms. The visual data mining takes the information visualization technology as a foundation, attempts to take advantage of our abilities of pattern recognition to make the judgement and the explanation to the mining outcomes, or makes the forecast to the trends of data. Several of data visualization techniques and methords are used to guide the whole process of data mining. These techniques and methords help the user to create two or three dimensional views about the data. By visually interaction with the views, the user could identify the intersecting information or pattern in the data.This dissertation firstly introduces the current development of visual data mining system, compares and analyses the merits and disadvantages of these systems. Secondly, the dissertation presents a prototype of a new visual data mining system. This system is devided into graph sub-system, data sub-system and network sub-system. Graph sub-system is used to visualize the data. Data sub-system indicates that a data mining system should have the ability of adapting complicated format data. Network sub-system supports communication within multi-users in real time. It is also used to test the validity of cooperative mining within multi-users.This dissertation concerns the design and implement of functional models of every sub-system in detail and presents one application based on every sub-system when it is finished to test its validity.