Wolfram|Alpha Computational Search Engine
This sounds like an interesting search engine project from the creator of Mathematica. It’s scheduled to go live on May 18.
The concept of the search engine is quite innovative: a Computational Search Engine, where it not only retrieves existing information from the web and various other data sets digested by the engine, but along with it it produce all sorts of useful computations, comparisons and charts.
Stephen Wolfram on his blog gives us some hints, talking about all the data and knowledge that’s know to mankind:
A lot of it is now on the web—in billions of pages of text. And with search engines, we can very efficiently search for specific terms and phrases in that text.
But we can’t compute from that. And in effect, we can only answer questions that have been literally asked before. We can look things up, but we can’t figure anything new out.
In this very impressive demo of the capabilities of the search engine, he also says:
We’re trying to take as much of the world knowledge as possible and make it computable.
Essentially, what Wolfram|Alpha promises to be is a search engine, powered by Wolfram’s own Mathematica, and containing all the qualitative and quantitative facts gathered by the creators of the engine, and plenty of algorithms to crunch these facts, compare and relate them to each other, and present the result in a human readable format.
Mathematica for the laymen
Actually, I highly recommend you go and watch this demo. It shows some amazing capabilities embedded in this search engine, accessible using simple queries in natural English. The results are impressive in their depth, relevance and presentation (of course, all of this will be put to the test soon). I really can’t add anything over this demo, if you have a few minutes, it’s worth your time. Sometimes, it’s even spooky. It makes me think of old scifi movies that feature an omniscient computer that can compute anything about anything with a click of a button or a simple command (think Ziggy from Quantum Leap for example, without the sultry voice — or it may be a feature coming soon?).
If it really launches with the capabilities demoed here, it’ll become a primary resource for every student, teacher, researcher or any sort of knowledge worker in the world. Moreover, its reach won’t be simply in academia, research, technology and similar fields, but I think it will be a hit among anyone with a bit of curiosity, and who has a question on his/her mind. All they have to do is ask it in plain English, and see what would come up… If this demo is an indication of what to come, plenty of interesting facts will come up. It’s essentially the power of Mathematica and Google (smartly) combined together, and put into the hands of laymen. Here’s a list of topics/data sets initially digested by Wolfram|Alfa:
Wolfram|Alpha Initial Knowledge Areas
Just a couple of examples to illustrate what all this means…
- Type “2 cups OJ” and it will produce all the nutritional information contained in two cups of Orange Juice.
- Type “gdp france” and it will produce a comprehensive display, including charts and time series data on the GDP of France.
- Type “gdp france italy” and it will produce the same as above, except that now it also compares the two countries together.
These are just some examples taken from the demo. Here’s the screenshot of the answer to the OJ question above:
Wolfram|Alpha results for "2 cups of OJ"
One interesting fact about Wolfram|Alfa is that all its data is “curated” by the creators of the search engine, i.e. by humans. This begs two questions:
- Are there automated mechanisms to update time-bound data that could go stale (such as stats of the GDP of France)? Are these mechanisms are human driven as well?
- What are the limits of such a system, given that it needs continuous input from its human “curators”? Is it going to reach a point where it’s too complicated and too hard to maintain and expand?
These questions may or may not be answered on May 18, when the system will launch. In all cases, mark your calendar. This promises to be quite fun.
Website: http://www.wolframalpha.com/
Wolfram Research is planning to broadcast the launch their new search engine Wolfram|Alpha live. This should be interesting to watch. It’s the first time anything of this sort is attempted. I’m curious what they will and will not show during this webcast. Some quotes from their blog:
we’ve been rather surprised that we haven’t been able to find even a single publicly available record of the commissioning of any large website at all
Some details of what will go on:
We’ll start webcasting our preparations at 7pm CDT (UTC -5). We’ll work through checklists, do a final test of our infrastructure—and then, if all goes well, within an hour or two we’ll have Wolfram|Alpha live on the public web for the first time
After Friday evening, we expect to be continuing to test and resolve infrastructure issues throughout the weekend
The webcast can be followed live here: http://www.justin.tv/wolframalpha
