Saturday, August 21, 2010

AGI on the WWW and graph visualization

If you have noticed, recently a lot of web sites have acquired Facebook's like button. At first I didn't see the broad implications of this tiny button if it wasn't for a f8 LIVE video I watched last week.

At this year's f8 LIVE, Mark Zuckerberg and Bret Taylor introduced the Open Graph protocol. For the total nubs, graphs are (according to Wikipedia's article on Graph theory) mathematical structures used to model pairwise relations between objects from a certain collection.

Now, I have read about graphs here and there. They're visible in the following fields:
  • My dictionaries' future projects all ended up in graph-like structures.
  • My recent reading of articles on AGI (Artificial General Intelligence) have also been related to graphs. If our gray matter is the synthesizing algorithm (see my article on the brain, in bulgarian) then it's cooperation with white matter (being a network layer) forms a good candidate for saying that our intelligence and consciousness are a decent implementation of a graph.
  • The NoSQL movement has made some graph databases sprout. I believe the slow development of technology and phat corporations like Oracle are amongst the reason why we haven't seen the boom of "not-relational"databases. Not until now!
  • The World Wide Web is also designed after this pattern. The Web 1.0 had pages. The Web 2.0 introduced applications. So now these applications want to communicate and process pages and data. Web 3.0 (Semantic Web) was always talked about but I didn't see it happening. RDF, OWL, these are mostly things of tech people and scientists. Now I think, practical technology has been developed and we're in need for that old idea of graphs.
What Facebook has done is basically open itself to the others. They have built and built and built, and now that there are nearly 500 million people hooked, they say "we're open for business". Or maybe they don't want to end up like MySpace :) For tech companies it's really important to be on the wave.

So here is Facebook's portfolio of new offerings:
  • Open Graph - Be part of the graph (FB will understand your web pages and build its graph)
  • Social Plugins - Personalised parts of FB on your website (human-readable FB graph)
  • Graph API - Program against FB's graph (machine-readable FB graph)
Now Facebook is all social, social, social, but what they're doing is just an implementation of something much more general. I guess they have grasped the people's needs and just like email and information on web pages was important for the Web 1.0, applications (social, e-commerce, etc.) for Web 2.0, now we can link all things and build the web of things/objects.

What is Facebook's idea? Build its graph (with the help of Open Graph or meta) and visualize it (Social Plugings, Graph API). They're the man-in-the-middle just like Google with its "graph" (with the help of Page Rank or links) and visualization (Search engine). I saw this Google relation somewhere but I don't remember well...

Of course the question of building a graph is important, but I'm a little more intrigued by the visualization. Do you need a lot of interfaces/applications/plugins to be able to display data? The information next to the Like button with just some icons and text. A representation of the data, let's say as a layer on Google Maps. I think we'll stick to this bunch of applications until we get the direct link to our brain :)

Now to the first part of this article's title. There is a lot of research done on AGI. What I see though, is that real world applications like Facebook and it's technology are coming from behind and bringing real intelligence to the Web. May it be that the evolution of the Web will bring us AGI and that it's the best place to develop AGI? I think I have to create my dictionary software on this new medium... graphs (neo4j?), Internet, people.

Thanks for reading, you two-three people :)

No comments:

Post a Comment