You want to know the current distance between the Earth and the moon, given that the moon's orbit is elliptical. Into a single field, you type the question as you would ask it of a human being -- and you get the answer. That's the vision of Wolfram Alpha, a new and potentially revolutionary "computational knowledge engine" that goes public next week.
Details are emerging about this latest creation from Stephen Wolfram, the Champaign, Ill.-based scientist whose Wolfram Research is best known for its Mathematica software and for A New Kind of Science (NKS), a best-selling book and approach to scientific problems using a set of basic formulas.
Built on Mathematica and NKS
On his company's blog, Wolfram has said his new answer tool was built using Mathematica's "symbolic language to represent anything," its "algorithmic power to do any kind of computation," and on NKS' "paradigm for understanding how all sorts of complexity could arise from simple rules."
Currently, he wrote, search engines are efficient in answering questions that have been asked before because they retrieve existing knowledge. But how do you ask a question about something that hasn't been answered before?
One way, according to Wolfram, could be the emerging semantic Web, where information is tagged as to its meaning. But another way, the approach taken in Wolfram Alpha, is using algorithms to "explicitly curate all data so that it is immediately computable."
If knowledge is computable, he said, natural-language inquiries don't need to be understood. They need to be represented in "a precise form that fits into the computations one can do."
In interviews with The New York Times and others, Wolfram is downplaying reports that Wolfram Alpha's target is search-engine king Google. As he has pointed out, his new tool doesn't search Web pages, but uses logic rules to interpret a huge knowledge base that is still growing.
The knowledge base, according to news reports, is being built by a Wolfram Research team of about 100 people. This kind of intelligent resource, if it works as hoped, has been a long-sought goal among computer-science researchers, especially in the field of artificial intelligence.
Google 'Doesn't Have to Worry' Yet
In opening up new ways to use computer-based knowledge, Wolfram Alpha could also open up major revenue streams, including advertising and other sponsorships, premium versions for specific uses, and leasing it to others, such as search engines.
Al Hilwa, program director at industry research firm IDC, said, "historically, it's been very tough to provide accurate searching" using natural-language interfaces, as compared to the keyword-based searches that Google and similar engines provide.
While noting that he hasn't seen Wolfram Alpha, Hilwa pointed out that the database is an even bigger nut to crack than the interface. "No database has been complete enough to satisfy any query," he said.
Hilwa suggested that Wolfram's target could be specific, science-based domains of knowledge, such as weather or the oil industry. Even with its specificity narrowed, he said, "it's hard to keep them up to date."
But that is a different territory than Google's main market. "I don't think Google has to worry about this just yet," Hilwa said. Nevertheless, Google is making moves into this area as well. Last month it announced that public data, and the ability to compare data, had been added to the popular engine.