In The News - July 2009
Wolfram|Alpha
On Tuesday, May 19th, Stephen Wolfram, creator of the computational software Mathematica, unveiled his latest project, a computational knowledge engine called Wolfram|Alpha. For those who have been paying attention, WolframAlpha has been described as Google's newest threat, but as it seems, our alarmist nature is running away with our common sense. Google is a referral service, a starting point in the search for information. Type in WWII, and Google will point you towards 18 million websites and wave you off with a thumbs-up and a "good luck!" Type WWII into WolframAlpha, and you get a very precise entry detailing dates, people and countries involved, and a timeline.
According to WolframAlpha, it's "the first step in an ambitious, long-term project to make all systematic knowledge immediately computable by anyone". How long is long-term? Well, since we are continually learning and adding to our accumulated knowledge, long-term, in this case, could end up being forever. To date, WolframAlpha boasts over 10 trillion pieces of data, over 50,000 types of algorithms and models, and linguistic capabilities for over 1000 domains. The numbers sound overwhelming and the possibilities exciting, but exploring reveals its potential as well as the limitations of this type of tool.
First of all, WolframAlpha is not a search engine. Instead of searching the web and returning links to be sorted through, WolframAlpha interprets your query, pulls together information from its knowledge base, and attempts to actually provide an answer. WolframAlpha deals with facts, and it can only know those facts that have been added to its database. While the knowledge base itself has an enormous capacity for growth, it is still limited by the rate at which information can be added. Fortunately, a growing community has sprung up around WolframAlpha, and their feedback has greatly contributed to the expansion and accuracy of WolframAlpha's knowledge base. Unlike information gathered from non-refereed sources that populate the internet, users can feel confident that information found on WolframAlpha is accurate and up-to-date.
For those users comfortable with Google, WolframAlpha will be a new and possibly frustrating experience. During its first week, WolframAlpha was able to provide answers to 75% of the 100 million queries posed, leaving approximately 25 million queries unanswered. One reason for this is that people are still learning the scope of the engine. WolframAlpha deals only with facts that are in the public domain. Subjective queries, such as "most dangerous mountain to climb", will get no response, but "tallest mountain" will return a list of mountains in order of elevation.
Another reason queries go unanswered has to do with the syntax of the queries themselves. Queries in the form of complete sentences can be answered, but WolframAlpha encourages users to employ only the most necessary search terms. The query "How do the GDPs of France and Canada compare?" will not result in an answer, but simply typing "GDP France Canada" will result in a page of charts and graphs comparing their annual GDPs dating as far back as 1970. For queries that receive no response, WolframAlpha gives users the option of clicking on "Tips for good results". In addition to avoiding complete sentences, these tips include avoiding abbreviations and making queries specific rather than general.
At first WolframAlpha seems very much like an encyclopedia, but as the "GDP France Canada" example illustrates, it is capable of synthesizing specific pieces of data to generate meaningful answers to complex queries. One neat thing WolframAlpha can do is take a series of musical notes and produce a scale you can actually listen to. Another useful feature is that WolframAlpha can give the exact nutritional content of the foods you eat. Type in "10 peanut M&Ms", and a nutrition label, similar to those you find on packaged food, appears letting you know that you've just eaten 103 calories, 5.2 grams of fat, and 2 grams of protein. WolframAlpha can also calculate mortgage payments and tell you what the weather was like on the day you were born.
WolframAlpha is capable of delivering quite a lot of information, if you know how and what to look for. For those who are interested in exploring WolframAlpha's possibilities, the home page provides examples from a number of topics such as chemistry, physics, and geography. These examples not only show what kind of information is available, they also show what the original query looks like. Paying attention to how queries are worded is a good way to find out how best to pose your own queries. For those who want more examples to study, click "more" under "Sample Topics" and a complete list of viable topics and subtopics will appear, along with a "Gallery of Examples" that continues to expand every time you click "more". If that's not enough, you can watch a 13 minute video tutorial conducted by the creator himself, Stephen Wolfram.
Since WolframAlpha is made possible by the computational software, Mathematica, one can immediately recognize how this tool might be especially exciting for mathematicians and math majors. Type in any mathematical equation, and behold with joy as the answer not only appears, it appears with all the steps worked out for you and any corresponding graphs.
Although it seems especially suited for math-based queries, WolframAlpha is not just a resource for math-minded folks. According to its homepage, the ultimate goal is to "make it possible to compute whatever can be computed about anything". In addition to music and nutrition (mentioned above) other subject areas that are already included in WolframAlpha's ever-growing knowledge base are astronomy, socioeconomic data, culture/media, education, sports/games, and money/finance. Type in "Mona Lisa" and WolframAlpha lets you know that the development of a visual art subject area is under investigation. Type in "Da Vinci", however, and you are presented with the artist's full name, place of birth, birth/death dates, and a brief timeline.
WolframAlpha even provides answers to some of the most probing questions ever posed, such as "what is the meaning of life" and "what is the airspeed velocity of an unladen swallow". It's good to know the developers are not afraid to tackle such questions, and I think it's especially important that despite the ambitious goals for WolframAlpha, the developers manage to maintain their sense of humor.
When one uses WolframAlpha for research purposes, it should be cited as a source. WolframAlpha claims to be a primary source, but the validity of that claim is still under review. Any professor requiring 3 primary sources for a paper may be extremely nonplussed to find WolframAlpha counted among that number. When in doubt, check with the professor when gathering sources. Primary source or not, WolframAlpha is a rich resource that people across disciplines can look to for answers and information.