this is a really cool concept at least when searching for people. After searching for my name it gave me a list of sites that have my name in them (usually mailing lists) and in the related email addresses section the first 3 email addresses were mine. The other two were people that I have talked to on some mailing list. The site is REALLY slow at the moment but I'm hoping these are issues they can work out.
Do they actually crawl the web for this or do they do some sort of searches in the background (it does take a long time to get the results back)?
I think it might be intentionally slow to encourage you to buy the pay/GUI version. They might also pay for their API calls in that one, there are a bunch of mirrors for this search engine because sometimes a particular instance doesn't work because it is out of search engine API calls for the day.
This works best as a security tool for "penetration testing" when you search for domains and IPs.
"Evolution works by taking a single node of data and applying a set of transforms to it, which result in a preselected number of new nodes being created."
From this paper, which contains no additional information about the node based approach. It does contain an image of the GUI version of the old version of Maltego, which looks like nodes.
http://www.milw0rm.com/papers/172
The only thing I've seen that mentions the algorithms is this powerpoint from a talk the program author gave, but it's clearly missing a lot without the actual speech he gave.
http://www.paterva.com/papers/CSW2007.ppt
Do they actually crawl the web for this or do they do some sort of searches in the background (it does take a long time to get the results back)?