Forum Moderators: open
It's currently under development, and there is an article and a beta of VisIT (Visualization of Information Tool) available for download at the University of Illinois site:
[news.uiuc.edu...]
Using its defaults, VisIT searches top ten results from Google and Fast and discards dupes. Gathers from each all the pages it can and creates mini-site maps for each. Fine line arrows show whatever cross-linking exists between all the results.
Mousing over any page icon delivers its meta tag or first-on-the-page text content. Clicking on any page launches DOS command.com which launches your native browser and opens the page.
Because it seeks to create these mini-site maps, the completion of a search takes a long time. After a few (3) searches, it ate up my RAM and began to make the cursor movement extremely choppy; once I exited the program, smooth mouse movement returned.
One of the foundation premises of this project (it is BETA), is that a site which receives links to it from its compatriot result sites is the best place to start to look for answers. That is, search for "blue widgets" and see who receives the most links from the others in this search result to find the best authority.
One feature which might be interesting to play around with some (other) day is the screen which enables you to re-set VisIT's default "weight preferences". In a half-dozen categories, you can increase the importance of inbound vs outbound links or the value of the order Google of Fast has placed on the site.
There are other features of VisIT which allow the searcher to save searches and cut and paste the results and mix with some stock graphics, but after an hour and a half of this, I, um, had enough and cannot report on these features.
Anyone who wants to explore further should keep some of these early issues in mind.
I was trying to think of how to do something just like that in html. hmmm
Floating layers with absolute positioning would work to a degree, but drawing lines would be a pain. Big imagemap would probably be easier.
And the new Visit, site:
[visit.uiuc.edu...]
Unfortunately I must have missed this thread the first time around. But it does seem like they made some improvements over the past 9 months. I installed it on a 266mhz NT machine and I'm not having any of the problems with ram or speed that you guys orginally reported.
Also, it looks like the list of search engines have changed. Your current options (you can choose which to query) are Fast, Excite, Yahoo, ODP, AV, Northern Light(not sure why they still have NL), and Inquiry.
This could be a very useful tool if your doing some heavy duty researching. I really like that they allow you to adjust the various weights for the algo variables.
In my opinion it would be little to none. The variables that they do allow you to control are not even close to what a real search engine would look at. They would need to make some major changes before this had any significant reverse engineering capabilities.
