Stephen Wolfram, a 52-year-old scientist, software designer and entrepreneur, tends to go his own way — often with noteworthy results. He published his first physics paper at 15, earned his Ph.D. from Caltech at 20 and two years later won a MacArthur prize.
Less than three years ago, Dr. Wolfram created a new kind of search engine, called Wolfram Alpha. Unlike Google or Microsoft’s Bing, Wolfram Alpha does not forage the Web. It culls its own painstakingly curated database to find answers.
There was skepticism in 2009, when Wolfram Alpha arrived, with critics saying the approach was very limited, useful mainly for math and science facts. But the technology has come a long way, including delivering many answers for Siri, the question-answering personal assistant in the Apple iPhone 4S.
The new version of Wolfram Alpha arrives Wednesday afternoon. Its formal name is Wolfram Alpha Pro, and Dr. Wolfram calls “Step 2, the next step of what can be done with this approach,” which he describes as a “computational knowledge engine.” This is a premium version of the search engine: $4.99 a month, or $2.99 for students.
The new version handles data and images. In a recent demonstration, Dr. Wolfram, using his computer mouse, dragged in a table of the gross domestic product figures for France for 1961 to 2010, and Wolfram Alpha produced on the Web page a color-coded bar chart, which could be downloaded in different document formats. He put in a table of campaign contributions to politicians over several years, and Wolfram Alpha generated a chart and brief summary, saying that House members received less on average than senators.
Dr. Wolfram dragged in a 3-D image and after a few seconds it rendered the image — a guitar — and reported the number of polygons (2,253), among other characteristics.
The Wolfram data-deciphering engine, however, was flummoxed by a table of occupational income figures plucked from the Bureau of Labor Statistics’ Web site. Dr. Wolfram suggested that it was confused by all the periods used to separate columns of numbers in the table.
This week is a beginning, Dr. Wolfram admitted. But, he added, “We’re starting to have the ability to understand data and images in the way we understand text queries.”
The text understanding of Wolfram Alpha has advanced steadily, with hundreds of subject domains added. They go well beyond the service’s origins, which built off the knowledge base in Mathematica, a popular math-formula software created by Dr. Wolfram. Serious math students, though, remain among Wolfram Alpha’s most avid users.
The subjects in the Wolfram Alpha database are now more useful to the average person. Type in “Tinker Tailor Soldier Spy showtimes,” and Wolfram Alpha delivers the schedule for local theaters. The movie times, Dr. Wolfram notes, come not from scouring the Web, but from a specialized information service.
Siri accounts for about a quarter of the queries fielded by Wolfram Alpha, whose staff has grown to 200. Several large companies in health care, financial services and oil and gas recently hired Dr. Wolfram’s private company, Wolfram Research, to do tailored corporate versions of Wolfram Alpha for them. Microsoft also licenses Wolfram Alpha technology.
Wolfram Alpha is one of a number of efforts to build greater understanding of the meaning of words — or semantics — into search, said Oren Etzioni, a computer scientist at the University of Washington. I.B.M.’s Watson, Apple’s Siri and Wolfram Alpha rely on structured databases of knowledge, while Google and Bing are trying to add more semantic understanding into general search engines.
The progress, Dr. Etzioni said, is good for the field. “It raises the stakes for everyone around the table,” he said.