Saturday, 2 May 2009

Wolfram Alpha: the next generation of Search?

The possibility of genuine semantic search being just around the corner has been around for several years. With, let's face it, the strongest contender being the bottomless R&D budget at Google HQ. So the very soft launch of Wolfram Alpha the other day (no product demonstration, lots of theory, a request for the tech bloggers invited to the launch webinar not to post screenshots, etc) shouldn't really be anything to get excited about. The product is not a search engine as such, rather a curated collection of data and statistics already available on the web. This data is organized using natural language processing in response to user search queries.

So not a competitor to the existing search behemoth, but potentially a radically different vertical search proposition. The caveat being that no-one outside of Wolfram programmers Mathematica, really knows what verticals, or how specialist.

However, two days after the demonstration, on the 1st of May, some screenshots started to appear, principally on ReadWriteWeb. These show the output from Alpha from research-orientated search queries:
"Internet Users in Europe"

So automatically outputs a range of the most relevant data relating to the query. As a contrast, I ran the same search in Google, and received this set of results :
"Internet Users in Europe"
There might even be some of the same data there, but to me the Wolfram results look much more usable: you wouldn't need to go any further with the query to have a reasonable set of answers. Google is more about giving you a starting point to look for more.

I don't think that this is in any way a 'Google-Killer', but I do think that I will get a huge amount of use out of the Wolfram system. I've been building my network and tag history on Delicious for a couple of years to get to a point where I can use Delicious as a starting point for most advertising/marketing/planning/digital related searches, but I also work in finance, travel, charity, youth, auto and sports sectors, and there isn't (as far as I have been able to discover) enough Delicious history or users in the UK to make these worth developing. NLP seems like the obvious successor to social search, and so I hope to be able to write up how I have been finding Wolfram in a month or so's time after the Beta launch.


neilperkin said...

I was talking to someone today who'd seen the demo and was raving about it - the key reason being (like you say) that useful information is presented without the need to click through. If it does this well enough, who knows, perhaps it might succeed where wikia / mahalo etc have not. Interesting prospect.

Post a Comment