Perhaps this uses DBpedia (http://wiki.dbpedia.org/About)?

This is similar to the site Freebase (http://www.freebase.com/), which has a lot more data. Example node: India (http://www.freebase.com/m/03rk0).

This definitely is based on Wikipedia categories, which can be extracted from DBpedia.

I suspect it uses DBPedia too. Me like :) This could really be something useful, personally Wolfram Alpha is way too noisy information-wise for me to use.

I'm not sure but it seems to use the category pages on the wikipedia pages to show you the tree.

This reminds me of "Nested" from Orteil, the creator of Cookie Clicker.

http://orteil.dashnet.org/nested

Could use another few iterations of the interface (I don't like the way the lists vanish right after they appear), but the concept and execution are first rate. I will be making regular use of this. Most semantic-web-like thing I've seen so far this year.

This site also summarizes what I hate in the current webdesign trends: broken back button and spurious event handlers attached to the not-quite-linky-links.

For almost all of my queries, this is just giving me the list of "categories" on the corresponding wikipedia page, leaving out the wiki-specific ones such as " Articles with Open Directory Project links", etc. Clicking the categories just opens the listings on the corresponding Wikipedia category page.

Quite possibly I don't understand what its doing or why, but I entered the word "stylus", meaning the needle on a turntable, and I got an opening suggestion of "writing implements", and the tree related only to that. So, unfortunately for me a bit of a fail. Like I say, perhaps I missed the point.

I assume it didn't have that meaning in whatever database it's using. Writing implements is a valid interpretation at least. It does seem to do disambiguation if you search for the classic 'pitch'.

It's just a fun way of walking from one concept to another, like browsing Wikipedia and ending up miles from where you started.

In fairness, the writing implement definition was around for a few thousand years before the invention of record players. Try 'vinyl stylus' - you're not restricted to single words.

with what i understand, the idea here is to basically start from somewhere, tread a path and get 'lost' somewhere else or into something. It's fun to walk through i guess.

What i liked is - now, i tried the word "stylus" like you did, but misspelled it as 'stylis'. it gave auto-correct suggestion - did you mean 'stylus'. good for a small site, i think.

You could maybe couple that with links found with wikipedia.org/wiki/*_(disambiguation) results

This is very cuil!

Nice. I thought Cuil had some interesting things they were doing.

Really impressed with results for words like 'twirling', 'homoiconicity', 'ferromagnetic', 'mahler'.

Unimpressed by 'sandwiches', which surprised me.

'sandwich' gives a great result. there is some song called sandwiches and it's taking that into context. may be looking for the closest match of the given word.

Annoyingly, the contextual menu makes it impossible to open the wikipedia articles in new tabs. One solution would to have actual links (either text or icons) after each title.

Interesting visualization -- perhaps having a link to the wikipedia page for each item or using an iframe to display the wikipedia page would make it a bit more useful.

alt+click is going to the wikipedia page

There are other platforms that don't have alt keys or where control click is a second mouse button.

wow. The best use i found is if you know something of a kind and want to find something like it.

Good start but a little shallow. Try 'Tolkien' or 'Lord of the Rings'...

Nice idea.

The "Explore " text is unnecessary spam. As is the fact that they are not real links.

I liked the idea. it's information travel.

Did we kill it? Not working for me in Firefox.

it's fun to browse through. I started with "India", ended up at Charlie Sheen ..

i like the way we could wander around :)

oh! well lots of data here. good one