Wolfram|Alpha is a new type of internet resource that has just gone “live”. Many are calling it a new type of “search engine”, which it technically is, but it isn’t a search engine in the way we are used to envisioning one. Others are calling it an “answer engine”, which isn’t a bad description. Wolfram takes the user’s query and builds a response from a variety of resources that resembles what one might get from an almanac or or encyclopedic resource. Wolfram’s own “about” page avoids any particular label.
Right now it has a relatively limited set of resources. It does well with towns, states, people, movies, word definitions and many scientific questions in the areas of mathematics and chemistry (which is understandable considering that it was built using Mathematica as the foundation. It doesn’t do well with music, books, television shows, and abstract concepts. Much of this has to do with the data sets and methodology that is in place, and I expect to see much greater depth and breadth represented as time goes on.
I have spent a bit of time last night and this afternoon playing with it, and am fairly impressed. Try a few queries, and be sure to view the “Source Information” link at the bottom of the results. This provides a fairly generic listing of the resources used for that type of query. Not all of the sources were necessarily used for your particluar query, but were the sources from which the data was extracted. One can see that with the addition of additional resources, Wolfram could become powerful first step for research.
Some sample queries:
- DeKalb, Illinois
- World According to Garp (movie)
- World According to Garp (novel) [note that this one fails… no resources yet for literature]
- Heaviest Gas Element [many variations of this don’t work, an example of syntax, I suppose]
and it doesn’t know quite how to handle leading articles, either:
This is going to be a great reference; it needs time to mature and for additional data sets to be included. This is certainly one place where the use of library cataloging information would be very, very beneficial. Imagine being able to connect the dots between books, authors, publication dates, settings/locales, etc. and other data sets.
There are many sources of data that are mostly silos waiting to be tapped, Infochimps.org being one example. How quickly and effectively they are able to incorporate useful data will partly determine how successful this resource becomes. Improving their context recognition (i.e. figuring out what a person wants to know based on their typed input) is the other, perhaps harder, challenge.
Other sources of commentary and information:
- Resource Shelf’s Brief Users Guide to Wolfram
- Quizzing the World’s First Answer Engine
- MetaFilter’s entry, and resulting discussion/commentary
- Wikipedia’s Wolfram Alpha entry
In closing, I like their nod to 2001, A Space Odyssey in their load exceeded error screen:
I'm sorry Dave, I'm afraid I can't do that...
