Automatic Visualization Doesnt Work

Many of the BioinformaticsCaseVultures? and UmlCaseVultures insist that you can automatically layout a 2 or 3-D diagram so that the human eye can automatically pick up meaningful relationships and associations.

But in fact you can't. The true topology of these subjects is far too complex to visualize in 2 or 3-D. So all your "visual insight" is just statistical projections and artifacts of the visualization - you get nothing that actually tells you where there are important distinctions, causes, or dynamics.

It is more appropriate to say computer aided visualization instead of "automatic visualization", since nothing happens really automatically. The human being always has to make decisions, especially in dealing with new data. Computer, however, can do a significant service in visualization. And to my opinion, there is large unexploited potential with computer aided visualization and knowledge discovery. -- JamesXinzhiLi

CategoryKnowledge