ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] [CG:] Wolfram Alpha is now available

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Duane Nickull <dnickull@xxxxxxxxx>
Date: Mon, 18 May 2009 06:47:45 -0700
Message-id: <C636B511.58D3%dnickull@xxxxxxxxx>
I have to agree with this.  In 1998 we launched something called “goXML”, what has been largely heralded as the first major contextual search engine taking advantage of structured XML markup.  While we never developed any ontology ourselves, the SDK allowed implementers to install a taxonomy or other hierarchic classification to alias terms.  You could effectively make declarations like

\\modular\transportationDevice\automobile <- \\car; \\vehicle;

Which would tell the results to handle any search for “car” or “vehicle” as if the searcher should find the first hierarchic classification.  More complex processing was allowed for terms with a plurality of meaning such as “washington”.  Searches could be filtered so the code processor knew to handle “washington” in a context of \\mammal\byProfession\Entertainer if the auxiliary term “Denzel” was used and different than if “Monument” or “george” were encountered.

We also had a natural language interface.  While rudimentary, this was more usable that the current wolfram alpha as it forced users to build sentences in a specific structure that could be less ambiguously mined.  The constructs were similar to “find me a” (other options included “locate”, “I want details about” etc.) <term> <predicate><taxonomic_term>.  Searches might be similar to “find me a horse that is rideable by a small child”.  

When we launched in 1998, we had over 51 unique patent points documented including aspects of the natural language interface, the contextual nature of the search and at least one new algorithm (the latter is almost unheard of).

The issue we encountered was that our index bloated heavily and grew exponentially larger than the content ingested.  In fact, at one time the equation was something like n ^ n + C where “C” was contexts.  We managed to reduce this substantially by rewriting it in C and porting the persistent storage to a hybrid B-tree and keyed lookup.

I highly suspect the engineers at WolframAlpha are having similar issues and realizing very quickly that launching a knowledge engine is a something that will make hardware salespeople very happy should it be fully funded.  Still – I hope it continues.  GoXML was arguably way to early to the marketplace.

Duane


On 5/18/09 6:20 AM, "Christopher Spottiswoode" <cms@xxxxxxxxxxxxx> wrote:

----- Original Message -----
From: "John F. Sowa" <sowa@xxxxxxxxxxx>

>Jon,
>
>That is the most frequent answer you get from it:
>
>> "Wolfram|Alpha isn't sure what to do with your input."
>
>That is why it is an alpha-level system.
>
>John

I'm afraid the problem cannot be solved along the standard
alpha-beta-production path.  In its very conception Wolfram|Alpha seems
unscalable:  do they really think they will be able to "curate" all the
world's computable knowledge?  Or direct all us other creators and
owners of knowledge into following their prescriptions?

On the other hand, Google also makes an analogous assumption, namely
that all the world's web pages can be handled by one system, their
system.  Wolfram is at least very up-front in basing his design on the
observation that the world's computable knowledge is limited.

But then again I'm afraid that his notion of "computational
irreducibility", which he explicitly (in his NKS anniversary blog of May
14) sees behind that limited scale of systematized knowledge, is a far
remove from it, so perhaps it is that gap that lies at the heart of that
unscalability?

Christopher

P.S.  How The Mainstream Architecture for Common Knowledge accommodates
that very same charge is a wonderful subject for another time...


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx



--
Sr. Technical Evangelist – Adobe Systems
Chair – OASIS SOA RM Technical Committee
Manager – Adobe LiveCycle ES Developers List
Blog: http://technoracle.blogspot.com
Twitter: duanechaos
TV Show: http://www.duanesworldtv.org
Band: http://www.myspace.com/22ndcentury

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (01)

<Prev in Thread] Current Thread [Next in Thread>