After months of anticipation, a new tool emerged this past week that should be of interest to reference librarians everywhere: Wolfram¦Alpha.
This new tool was designed by Stephen Wolfram, the wunderkind who developed Mathematica. Wolfram Alpha is assuredly not a search engine, even though it looks like one. It’s more an answer engine or a knowledge engine. Wolfram has taken a huge amount of data about the world--geography stuff, economic stuff, name databases--just tons and tons and tons of facts. Then they linked it all using the Mathematica language, so that the system itself can make inferences for you that you may not have thought to ask about. They are using the open web as a part of that knowledge base, but from interviews it sounds like the majority of the information that is being crunched is curated by the Wolfram team.
The most useful way to try and learn what the tool can do is to browse the Sample Topics located near the bottom of the page, or try a few of the sample searches on the side. It takes a while to get used to how it works, since you want to treat it like a search engine or a database, but, again, it’s not one.
Some of Wolfram Alpha's best features: On the results page, you can actually click “sources” and see the sources that the engine used to get the answer that it gave you. You can also get a PDF of the results that looks great, and could be easily slipped into a report.
Wolfram Alpha does a great job of comparing one entity to another: one country vs. another, one stock vs. another,etc. The engine will answer ANY math problem, as well--you can just copy/paste the formula in, and solve advanced math, including graphing.
Here’s an example of the sort of infrerences that Wolfram Alpha can make: if you search for your first name, not only will it give you the popularity of your name, and the years it was most popular, but it will also figure out approximately how old you are likely to be based on those two facts. Wolfram Alpha didn't find that fact somewhere, it analyzed the sets of data it has and "discovered" the correlation.
Give it a try…it’s likely to be a really useful tool in the reference world over the next several years.