Skip to content

raindropper

pouring…

Category Archives: Intelligence

When we consider the world around us, with all its images, sounds and concepts, traditional programming languages seem terribly limited. Look around you, and see how far from reality manipulating information like this is:

a = a + 1;

if (a > 23) then a = 0;

Of course, the programming languages reflect the architecture of the machine they are executed on.

What I’m looking is for a way of programming which enables systems have characteristics like these:

– Creativity. Show the system a chair and a human being sitting on it. The system generates and presents thousands of alternative versions of a chair. All in 3D, of course. Real world is 3D so no reason to aim any lower than that.

– Curiosity. The system wants to learn more.

– Photographic memory. WYSIWYP, what you see is what you process. Visual information provided to the system stays in its memory and it can manipulate it at will.

– Shades of gray. Things are not only black or white, true or false, 1 or 0, they can be something in between.

Actually, this seems not to be so about programming languages. This is about artificial intelligence.

[ to be continued & revised ]

 

 

 

 

 

Advertisements

Ok, Wolfram|Alpha – the computational knowledge engine – is up and running. Granted, I’ve seen three different versions of “site is currently under heavy load”. Do remember that there’s a large amount of on the fly computation and algorithm crunching behind each result. 

These are the queries I tested it with:

Where are you? gives a good result.

Where am I? gives the correct answer.

Who am I? gives the same answer.

sin(n/10) * 100 draws a nice chart.

What are you doing? is the first computed tweet. 🙂

What time is it? surprisingly gives nothing.

green, redminimum result.

BMW presents stock information.

Weather in Lahti Wow!

Weather in Lahti June 2003 Double-Wow!

Neuron is not that interesting for Wolfram to have knowledge about.

Are you OK? opens a “Human Discourse” functionality which is under development. What will it be?

All in all, Wolfram|Alpha provides an interesting approach and implementation. It’s certainly one to follow and use.

However, I do get “everything is a number or a taxonomy” feel from the data it contains. It mostly answers with numbers; even Madonna boils down to a straight line between two dates.

So, Wolfram|Alpha is the engine Douglas Adams wrote about

Google left, Wolfram|Alpha right.

Google left, Wolfram|Alpha right.

I do not know why, but the system creates an impression of an autistic Rain Man recalling phone book numbers and curated minutia with precision. Whereas Google is the outgoing guy with all the fun; its bots gathering data from the web carelessly, and giving noisy, vague answers at times.

Crystal Set Radio - Neuron

… Learnings from The “On How The Brain Functions” Experiment.

You, the web wanderer, are perhaps aware that a bit over one year ago I posted my research paper about theory of how the brain functions. I also built a blog around it so that you, the brain researcher, can comment it with witty references to tin-foil hats. 😉

The current gain is zero remarks to tin-foil hats.

Seriously, though, I still stand behind the theory and see several strenghts in it. The question is: why I’m not building a working prototype as it only takes a few diodes, capacitors and coils to make it. Maybe it’s because I do not want to disturb the local neighbourhood with electromagnetic noise.  

And: I’d like to thank Joni Tuoreniemi and Paul Tudsbury for commenting it and creating conversation. Thank you!

Sig’s posts on Thingamy and RDF triples inspired me to “put the verb” back to this interlinked mass of documents we call the Web. In a simplified way. 

Here’s the official W3C recommendation and specification of links: 

http://www.w3.org/TR/html401/struct/links.html

If you go through it, you perhaps end up in the same conclusion with me: There’s no simple way to define a semantic relationship between web resources.

I propose that we extend the A HREF tag so that it carries more meaning with it. Let’s borrow the PREDICATE concept from RDF and add that to A HREF tag. 

Here’s an example. On nokia.com, there could be a link like this: 

<A HREF=”www.nseries.com” PRED=”produces, sells, markets”>Nseries</A>

So, what would we achieve with this? It would be a way to provide more precise information, more knowledge for the search engines, etc… We would be closer to the Internet Singularity

Some of you have already heard about W3C’s Semantic Web framework. Tim-Berners Lee, the inventor of the Web,  presented roadmap for Semantic Web in 1998.

The framework can be summarized as:

“The Semantic Web is a web of data.”

“The Semantic Web is about two things. It is about common formats for integration and combination of data drawn from diverse sources, where on the original Web mainly concentrated on the interchange of documents. It is also about language for recording how the data relates to real world objects.”

Web today is about sharing documents, Semantic Web is about sharing and reusing data. If it gets the momentum, it will break the boundaries between applications, free the data from their silos. W3C provides a concrete example: my calendar application could – if I wanted – display my bank transactions from this week, and also the photos I’ve taken, day by day.

Some of the tools that the Semantic Web will be powered by are RDF, a knowledge modeling language, and SPARQL, a query language for RDF.

RDF is a method of describing data and resources formally so that they become accessible and understandable for software. Here‘s a primer for it. Berners-Lee’s view is that the future lies in “programming at the RDF level”.  

Below is an example of a SPARQL query. It displays title and price for books that are priced below 30.5. 

PREFIX dc: <http://purl.org/dc/elements/1.1/&gt;
PREFIX ns: <http://example.org/ns#&gt;
SELECT ?title ?price
WHERE { ?x ns:price ?price .
  FILTER (?price < 30.5) .
  ?x dc:title ?title . }

On IBM’s developerWorks interview (28.7.2006), Berners-Lee gives his view on the status of Semantic Web:

“I hope the Semantic Web will take off so that the data basically all the data which is out there which you have access to, to the Web pages, will now be available as data so you can treat it as data. There will be lots of very exciting applications built on that.  And we’re starting to see that now, but it really is, you know, we’re seriously into the exponential growth of the Semantic Web right now, and that’s very exciting.”

Some questions arise in my mind:

Is the original Web – this Web here you’re using now – non-semantic?

Where’s the boundary between data and a document? If Semantic Web uses RDF documents, will it eventually fall to the same “trap” with the original Web?    

Considering the massive amount of data in the Web today, could it be somehow utilized and reused in building the Semantic Web?   

Is there a demo somewhere showing the power of Semantic Web?

Do we really need another query language with its own syntax – why not expand SQL or use English? Could this query be enough:

book title, book price, book price < 30.5

Gary Flake runs Microsoft Live Labs, MS Research unit which is concentrated solely to the Internet.

His one-year-old presentation “How I Learned to Stop Worrying and Love the Imminent Internet Singularity” indicates the Road Ahead for Microsoft’s Internet strategy. 

When I hear the word “singularity”, it brings an image to my mind of all things united to one place, then Boom!!! – and a silence that lasts forever…

IMO Microsoft has been quite silent about Artificial Intelligence in the past. But now it seems that they are really starting to focus on that, utilizing the massive amount of data in the web and its feedback mechanisms.

As an experiment I decided to open a dedicated blog for gathering feedback and comments for “On How The Brain Functions” article. So, take a deep breath and jump to here.

I wish You a Marvellous New Year 2007!