Wolfram|Alpha isn’t sure what to do with your input.

May 16, 2009


Well, if you have lived the last days on a cage, maybe you haven’t heard of Wolfram|Alpha.

People say it’s the latest Google Killer, but it doesn’t do the same thing as Google. Wolfram|Alpha targets to answer a query with information, knowledge, rather than with web pages. This means that if you search for a date (let’s say, ‘june 23, 1988‘), it should tell you about the day (duration of daylight, phase of the moon, important facts on this date, and any other data it finds useful), instead of getting you to a Wikipedia page that potentially have this information.

The main differences with Google (and 99,99% of the rest of the search engines) is the presentation (it shows useful data to you, not a page where you have to find the information) and the query reasoner (it shouldn’t work in a keyword-match basis). This is all we could want: You type what you want (in natural language) and the engine shows what you want to see (real information, real knowledge).

The news are great, I know. Wolfram|Alpha could change the world as we know. And maybe it has an incredible value for those who are looking for specific information about one source (generic information about Rio de Janeiro, for example). Or if you want to know something about some kind of data you know Wolfram|Alpha supports (they have TONS of input templates and functions that you can use. You can probably stay all night up just checking a small subset of the input types they support. Try Web and Computer Systems, for example).

But even with this range of valid inputs, it can never figure out what I’m talking about. This post’s title is the phrase I’m reading the most since I began testing this engine. If I look for ‘mexican restaurants in rio de janeiro’, I would like to see an automatic mashup, using google maps and live information on the web, locating all mexican restaurants in rio de janeiro. Right now, they simply cannot figure out what I want. It doesn’t know of ‘Brazil’s population growth rate’ either, and many others that I tried.

Most of this information can probably be retrieved from Wolfram|Alpha using better formed queries. But I don’t think web people are willing to learn how to make queries all over again. Don’t get me wrong: I’m not telling Wolfram|Alpha isn’t great. They already got extremely good results (and got around extremely difficult problems), but I think they have much to work on input pattern recognition and integration, to make the service useful for the standard web user. There is much work left for “making the world knowledge computable” (and available), as this is their objective.

I’ll be posting here some more queries I made that returned nothing (and others that returned great results), do you have some too? Any thoughts about the service? Post it on the comments!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: