Wolfram Alpha 'Knowledge Engine' is Like a Modern Farmer's Almanac

May 8, 2009 by Lisa Zyga weblog
wolfram alpha
Wolfram Alpha is easy to stump, but still computes a wealth of raw data.

(PhysOrg.com) -- Currently, there's a lot of hype and skepticism surrounding the latest "Google rival," a so-called search engine named Wolfram Alpha. In the near future, anyone with Internet access will be able to freely visit www.wolframalpha.com and investigate how it works. In a media webinar earlier today, Wolfram Alpha's creator, Stephen Wolfram, said the site should be launching "in a bit over a week." Wolfram also showed off his ambitious project by demonstrating a variety of search queries and answering questions from journalists.

First, the question everyone is wondering: Wolfram Alpha doesn't know everything, and in fact it's pretty easy to stump it. Unfortunately, many users will probably try to confuse Alpha, and miss out on what it actually can do - which is provide tons of verified data and comparative information, prepared by experts and displayed freely and conveniently for the individual user.

Another big question is, will Alpha out-compete Google? Judging from the demonstration, the answer is definitely not - not anymore than Wikipedia has cut into Google's territory, at least. Alpha is like a Wikipedia of pure, raw data, with almost no word sentences, but full of charts, graphs, tables, maps, figures, etc. A bit nerdy-sounding, but on the whole, it's quite clearly displayed and easy to skim and interpret. As most people who use the Internet a lot know, there are times when you search , and times when you search or YouTube - and soon there will probably be times when you know that the type of information you want is more appropriate for an Alpha query.

As Wolfram explained, Wolfram Alpha is a "computable almanac." Like the famous Farmer's Almanac, it can display lots of statistics (that sometimes seem irrelevant from a practical perspective), but can be interesting to peruse, and often essential for research purposes. However, unlike a paper almanac, Wolfram Alpha doesn't store data, it computes data. It's not a , it's a "computational knowledge engine." It deals with systematic knowledge, computing new data from algorithms and generating true facts. In other words, it seems to be less human, and more computer.

Wolfram breaks the concept down into four pillars. First, data curation - data was taken from public and licensed sources, validated by human domain experts, and cited in a link at the bottom of each page called "sources." Second, computational algorithms - developers encoded methods, models, and equations in a form that can be computed, using 5-6 million lines of Mathematica code. Third, linguistic processing - mapping natural language words into symbolic representations that can be computed. Fourth, automated presentation - presenting the results in a way that's useful to people.

These four components are each very challenging, and, as Wolfram explained, far from complete. Alpha is a long-term project, and the company plans to continue working on it and improving on it for years in the future. Wolfram said that Alpha is "built on a Mathematica platform and on A New Kind of Science (ANKS) paradigm" (Wolfram's tome on the nature of computation). Here, simple rules lead to great complexity, and the goal is that all knowledge might be represented by simple rules.

In today's webinar, Wolfram demonstrated many search queries on Alpha (although screen shots were not allowed). For example, "gdp france" generated plots and graphs, with each new set of data displayed in its own module or pod. Then, Wolfram typed "what is the gdp of france / spain" and the program computed a comparison of both countries, with overlapping data in the same chart. (Interestingly, this example is somewhat similar to Google's latest ability of generating synthesized graphs for data such as unemployment rates in multiple cities.)

Other diverse and noteworthy examples include "president of argentina in 1943," which resulted in Ramon S. Castillo; "tide NYC 11/6/2020" which displayed a graph of the tide in New York City that day; "15 flips 10 heads" which generated probability and distribution charts of such an event; "ISS" which displayed a map of the location of the International Space Station in real time; "1.343495843" which computed the possible formulas that give you that particular number; "a__n__g" which gave words that could fill in the blanks of a crossword puzzle (amending and atoning); "earthquakes" which generated a global map highlighting earthquake locations in the past 24 hours; "2 cups oj 1 slice cheddar cheese," which computed a synthesized nutrition label of the snack including vitamins and minerals; and "running 4 mph 30 minutes age 40 male 5'8" 160 lbs" which presented charts of calories, fat burned, heart rate, and race prediction times for different distances.

Just how easy is it to stump Alpha? Journalists' queries that confused the engine included "nutmeg production USA," "teenage pregnancy USA," "area of a soccer pitch" (although it could do "area of a football field"), and, not surprisingly, "what type of sunglasses was Justin Timberlake wearing at the Oscars" (though there was a brief entry on the singer himself). For such queries, users receive the answer "Wolfram Alpha isn't sure what to do with your input."

Wolfram emphasized that the project was a work in progress. Among other things, he hopes to move past an English-only version. When asked why he kept it a secret for so long, he said that he was not sure that it was going to work. He said he had hoped to do "a nice, quiet, soft launch," but it seems more people were interested in the latest Mathematica-backed endeavor than he (admittedly) realized.

Wolfram added that, besides the free version, there will be a professional version for corporations that want to customize the program. While the free version will generate PDFs of the pages, the professional version could export data to spreadsheets, allow users to import their own data and use Alpha's computation abilities to analyze it, and allow users to store their own data and share it within the company. There will also be a method for people to contribute facts, which is not yet available.

© 2009 PhysOrg.com

Explore further: Wolfram Alpha Could Answer Questions that Google Can't

Related Stories

Wolfram Alpha Could Answer Questions that Google Can't

March 9, 2009

(PhysOrg.com) -- A new search engine described as an "electronic brain" could make searching the Internet more intelligent. Called Wolfram Alpha, the search engine computes its own answers rather than looking them up in a ...

Mathematica Users Get 100x Performance Boost From NVIDIA CUDA

November 18, 2008

At SC08, Wolfram Research will demonstrate a new version of Mathematica, the world’s most powerful general computational software, that integrates CUDA, NVIDIA’s parallel GPU computing architecture. This new version is ...

Google Releases Chrome 2.0 Alpha

January 14, 2009

(PhysOrg.com) -- With the full release of Chrome 1.0 in December, Google has just released Chrome 2.0 alpha that brings many noticeable improvements over Chrome 1.0. With this new alpha release of Chrome 2.0, the browser ...

Google search gets semantic

March 24, 2009

Google on Tuesday modified its globally popular Internet search service to understand relationships between words, as the company bids to better grasp what Web users are looking for.

No More HP Alpha Processors

August 19, 2004

Hewlett-Packard will release its final Alpha processor on Monday - the 1.3GHZ Alpha EV7z. The chip will be incorporated into the existing HP GS 1280 server. HP will continue to sell its AlphaServers through 2006, and will ...

Google Marches Toward Universal Search

May 17, 2007

Beginning today, Google is starting down a path toward integrating search information from previously disparate sources like books, Web sites, video, maps, news, and more.

Recommended for you

Making AI systems that see the world as humans do

January 19, 2017

A Northwestern University team developed a new computational model that performs at human levels on a standard intelligence test. This work is an important step toward making artificial intelligence systems that see and understand ...

Firms push hydrogen as top green energy source

January 18, 2017

Over a dozen leading European and Asian firms have teamed up to promote the use of hydrogen as a clean fuel and cut the production of harmful gasses that lead to global warming.


Adjust slider to filter visible comments by rank

Display comments: newest first

5 / 5 (2) May 08, 2009
kudos, thank you mr. wolfram. you are one of the few out there, who was acknowledged during your time for some obvious achievements, and who has non-obvious achievements which will continue to be explored far longer than your time.
not rated yet May 08, 2009
I'm not sure about: "However, unlike a paper almanac, Wolfram Alpha doesn't store data, it computes data."

Surely it DOES store data, computes data, and outputs INFORMATION.
not rated yet May 08, 2009
Many people in the academia envy Wolfram.

Impressive that he keeps going at a good rate with awesome ideas. Congrats for this product, thank you.
not rated yet May 09, 2009
There will be people who find this usefull, but i guess not many people are going to contribute when functions like exporting to spreadsheet are pay-only.
not rated yet May 09, 2009
Its nice to finaly have a serch engine that isn't dependant on marketing. This is going right to the top of my favorites list.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.