Wolfram Alpha coming - beware Google

ahu

Newcomer
It's not a dumb search engine like Google. It's not based on semantic web technologies either. It's an immensely ambitious project. Wolfram stating it's more ambitious than Wolfram's earlier ventures Mathematica and NKS says a lot :D

Take a look at Wolfram's blog:
http://blog.wolfram.com/2009/03/05/wolframalpha-is-coming/

I think this is not going compete with the main Google target audience though. It will however be big...
 
Jon Honeyball goes a bit weak-kneed about Wolfram Alpha in his Epilog column in the latest edition of PCPro magazine here in the UK. A lot of the article is in praise of Wolfram as a company (and moderately critical of Microsoft's corporate inertia) but, as for the Wolfram Alpha tech itself, Honeyball says he was "blown away" by the "stunning" demonstrations and heartily recommends that everyone try it out once the site goes live.

Sounds interesting and it would be nice to be able to search for things without having to sift through pages of Google results to find something useful as is so often the case.
 
I'll hold judgment until I see the released product. I love Mathematica, but some of the things he's talking about don't quite fit together.

For example, he's talking about how they simply "aren't there yet" for processing natural language, but somehow he's also talking about making all the online materials on the internet into computable data? If it's computable but not in a natural language way, then how is my mom going to interact with it?

Maybe the answer is simply that she won't, and this search engine will be targeted to people who can grasp the syntax and concept. But that doesn't seem like a world-beater search engine, either...
 
I'll hold judgment until I see the released product. I love Mathematica, but some of the things he's talking about don't quite fit together.

For example, he's talking about how they simply "aren't there yet" for processing natural language, but somehow he's also talking about making all the online materials on the internet into computable data? If it's computable but not in a natural language way, then how is my mom going to interact with it?

AFAIK the emphasis isn't on parsing exact English grammar. You wouldn't want to use the thing like that anyway, rather one would type the relevant keywords or maybe use some abbreviations. The engine then tries to parse the input into objects in the system's curated knowledge base and finally compute the answer using Mathematica-type algorithms.

The computable aspect of the query engine is interesting, as the system seems to use the NKS-paradigm on several levels, as revealed in an interview by Rudy Rucker. It might even find new algorithms and solutions from the computational universe.
 
It was pretty much overhyped bogus from what I read about it ages ago. It isn't exactly like this has not been talked about before. I suppose someone has to stoke the hype in forums though.
 
So how is this thing supposed to work?

My understanding is that you ask it a question and it basically tells you an answer, rather than pointing you to links that may contain the answer you're looking for.

I have a lot of questions about how the "curate" their data, and how they deal with intentionally incorrect information. Would someone have to comb through each piece of data and check for accuracy? That would never be possible. Maybe it would return answers in a citation format? Like, it would return an answer addressing a multitude of possible answers and have citations for each piece of information so you could do your own fact checking.

Pretty interesting, and I'm looking forward to seeing it in action, but I don't see how this could ever be built or maintained in a practical way. For instance, with medical information, there are constantly studies that suggest something may be true, and within a few years you'll have other studies that publish contradictory information, or information that disproves the first study. To make all the connections and weight the answer to suggest the most recent or least biased information would be impossible.

One thing I find interesting would be the use of this as a customer service tool. For instance, I could go to the Dell website and I could use something like this to address customer support, rather than digging through forums, FAQs and guides.
 
I agree with all of your points Scott, and I especially like this one
One thing I find interesting would be the use of this as a customer service tool. For instance, I could go to the Dell website and I could use something like this to address customer support, rather than digging through forums, FAQs and guides.
I think you've probably got the best idea I've seen presented thus far; this would be far more useful application of the technology IMO.
 
I agree with all of your points Scott, and I especially like this one

I think you've probably got the best idea I've seen presented thus far; this would be far more useful application of the technology IMO.
First someone has to show that it can be done at all, after which many others try to copy it. One of them might make such a consumer service tool.
 
It's now live and I'm happy to report that it really works! And it works really well, apart from being slow (this should be fixed monday).

I have to say it's been some time since I've been so excited about a software product. It does have it's limitations (sorry Davros;)) and it's not for everybody.

The thing I'm a little worried is how are they going to handle the enormous computing requirements when this really takes off globally.
 
Pretty impressive stuff! Not a search engine per se, but a really great way to get information.
 
The biggest problem is going to be the lack of 'curated' information. There are many areas where the engine just has no information available at present. How long before this stuff is added? I'd imagine you could have a whole raft of people working endlessly to add information without getting close to filling the many gaps.
 
yeah I tried searching 'german military production in world war two' and it didn't know what to do with the string. I would have thought that'd be a pretty easy question for alpha. I think this product is going to die really fast.
 
Back
Top