Wolfram Alpha Computational Knowledge Engine – First Official Demo

Posted on April 30, 2009
Filed Under collective intelligence, creativity, online search, online tools, online video, search engine, semantic web, web evolution | View Comments

Two days ago Stephen Wolfram gave an early preview of his “computational knowledge engine” Wolfram Alpha at a talk at Harvard University.

The video of the whole presentation (1h 45min long) is here above, while down below you can find some highlights I transcripted from his speech. Unfortunately we can’t see the screen of the demo during the presentation.

Also, today he’s been writing further about the project on his blog.

Basically the aim of the Alpha project is to find a way to make the systematic knowledge we’ve been accumulating over the ages computable.

The two big projects Wolfram worked on in the past years have made this possible. “Mathematica” (he’s worked on it for 23 years) makes it possible to do complex math and symbolic language manipulation. “A New Kind of Science” (NKS) has made it possible to understand much about the world computationally, often with very simple rules.

So now we have Wolfram Alpha using a New Kind of Science principles and the Mathematica engine.
The objective is to reach an expert level knowledge across a very wide range of domains.

If Alpha is truly as impressive as it looks in the demo, and it can handle the scalability challenge of public availability, then this is a huge advance on the state of online analytical tools.

A mathematical engine behind the scenes makes possible a much wider range of inquiries than the search-and-retrieve model provided by traditional search engines and data visualization services.

Some highlights from the presentation:

The goal is to define a way to make the systematic knowledge that we’ve been accumulating in our civilization computable; to find a way to take out all that data and methods and models that we know and somehow combine them to be able to compute whatever can be computed about the world.

Two previous projects – NKS and Mathematica – have made this not completely crazy.

I worked on Mathematica for 23 years, a platform for computing formal things, and for representing things formally.
A powerful sort of symbolic language for representing things, and providing algorithms about things.

The other project has been A New Kind Of Science, for studying the computational universe of possible programs, and what they do. NKS showed how it possible to understand most things of the world computationally, and more important that there are often incredible simple rules that lead to incredible complexities and richness.

So I started to think that all that knowledge out there might be representable, and handleable by rules that are simple enough that one can actually build them.
That’s what we’ve been trying yo do, with lots of practical ideas from NKS, built on the platform that Mathematica provides.

The first instance of this technology that we’re realising is the website called Wolfram Alpha, launching in a few weeks (mid-May).

The basic idea is: you type in a question, and WA type back an answer.

Examples shall include: GDP (gross domestic product) of France;the number of Internet users in Europe; the weather today in Lexington, Massachussets; and so on.

WA is concentrating on giving us facts about things, for more narrative content you can get Wikipedia links next to the results.

It allows to discover things as well, for example by typing “4000 words” we get a variety of associated items with that quantity of words, like essays.

Other sources include medical or financial data.
Or you can ask WA to compute “red plus yellow”, “the musical scale C”, how many people have PhD in Boston, or just type “orange juice” and get nutrition tables in comparison with other foods…
or even very precise, located and articulated questions, as “weather in Lexington the day Kurt Godel died”.

As Stephen Wolfram wrote on his blog yesterday, we’re talking about a really ambitious project, inspired by Leibniz idea of “finding a way to mechanize the resolution of all human arguments“, using logic and mathematics on a machine.

Sources: Wolfram Alpha Blog
Berkman Center for Internet & Society at Harvard University
Open Analysis