Research On Som-based Web Minner And Visualization Of It's Result

The continuous development of internet technology makes people easy to collect, store and manage knowledge, but people's ability of understanding knowledge has not significantly improved in term of the behavioral habits, thinking ways.In the "information explosion" age, the amount of information people got has been large enough, but how to get the most effective information they need, filter spam and how to read such a large amount of information without missing, all these become a big problem in front of us.These are problems named "information trap", "Information Trek," and so on they are all cause by "information explosion". In response to these problems, experts and scholars in and out of our country have made a lot of theories and methods; the most important thing is to do knowledge organization and knowledge visualization work.This thesis bases on the study of the present research situation on clustering results of text mining, and in the text mining system, it introduces SOM neural network algorithm, which has a good effect on visualization of knowledge.However, simply copying the SOM algorithm for clustering text mining will arise efficiency problems, for that the information on the Internet is exponentially increasing, and the results of textmining are increasing too,these will lead to a serious decline of efficiency, Therefore, the algorithm must make certain improvements. We use the defense ontology vocabulary as the input neurons and the output template of SOM clustering results, clustering step by step makes cluster more effective.It will well collect defense information and translate into color blocks chart, line graph, column chart, pie charts, three-dimensional column chart ,three-dimensional pie chart to show information to the user.Experimental results show that: this method of clustering results is accurate and more efficient to solve large sample of text mining under the SOM clustering.Besides, visual display interface with a tree diagram looks simple, users feel better.And it is very convenient for user to understand hot topic, gain knowledge, and make the decision.