Re: [ANN] A pragmatic implementation of the Semantic Web

Hi Danny,



On Apr 8, 2008, at 11:41 AM, Danny Ayers wrote:

> On 08/04/2008, ALT Mobile DEV <dev@altmobile.com> wrote:
>
> Hi Dan,
>
> ...
>
> [Zaid] You eschew discourse by discouraging others to investigate. I  
> tell you that we've solved some of the fundamental challenges of the  
> SemWeb and you turn a blind eye and refuse to look. I tell you that  
> we are in sync with TimBL's updated suggestions for a SemWeb  
> implementation and you cover your ears.
>
>
> While I generally agree with Dan's comments, I wasn't discouraged to  
> investigate. I had a look around your site, skimmed some docs and  
> followed the download link: XML Studio. The application failed on  
> startup, so I emailed your support address.

[Zaid] It's a well known Ubuntu 7 incompatibility with Java. We've  
been shipping for Mac, XP, Vista, Red Hat, and Solaris for years...  
but Ubuntu and Suse are a bit new to use. Here's a link which talks  
about the problem

https://answers.launchpad.net/ubuntu/+question/10270

This has been solved for the most part by Ubuntu and Sun in most  
system configurations.



>
>
> The docs on your site had a similar tone to your intro mail - "we've  
> fixed these problems" - but there was little explanation of the  
> perceived problems or fixes.


[Zaid] the problem is that the W3C Semantic Web standards do not  
_today_ provide the ability for users to make use of meta data that is  
relevant to them on normal web pages. The W3C SemWeb standards _today_  
only prescribe a solution for publishers. WWW users are excluded from  
the meta data revolution unless a publisher spoon feeds them. That's  
why nobody trusts META element-defined meta data.

In our implementation of the SemWeb vision for meta data, we allow  
users via Mashup technologies to generate meta data automatically.  
Here's a screenshot

http://java.sun.com/products/jfc/tsc/sightings/S15/Mobile/StudioScreenSnapz002.jpg


of our Mashup tool allowing a user to extract an HTML TABLE of Wi-Fi  
coffee shops in San Jose. The user just wants the list of coffee shops  
and not the ads and not the Dublin Core information. We generate meta  
data that describes the list of coffee places and a speech synthesis  
version of that TABLE is included just in case that Mashup user is:

1. vision impaired
2. a non-American English speaker
3. a power user who wants to listen to his content


TimBL discusses these type of issues when he talks about the "Data  
Web" and Mashups.. where users access HTML elements rather than HTML  
documents.


We have faithfully attempted to implement the TimBL "Data Web" and the  
TimO  and JonU  "the web as a database" concepts using our Mashup  
technologies.


> There seemed to be a big emphasis on (HTML) document metadata,  
> certainly an important part of semweb activities, but not the whole  
> story, a lot of the material we're dealing with is plain old- 
> fashioned data.

[ZAID] Yes, and while the W3C SemWeb recent standards are focussed on  
XHTML, we are announcing that we think that we've solved the HTML meta  
data part. But to do so, we had to shun the W3C publisher-exclusive  
meta data approach and allow WWW users control of web data and meta  
data. That's why we call this: W3C vs WWW. The meta data standards for  
the last 10 years are about the publisher only. By including the user- 
centeredness of Web 2.0, we unify meta data support for all WWW users  
including readers, writers, and intermediaries including search engine  
vendors.


> There was also a fair bit about client-side mashups, and around the  
> Web 2.0 world they are things that can be very useful and go some  
> way to suggest the potential of Semantic Web technologies (and if  
> the relevant standards are used, mashups generally become trivial  
> because you don't need to build a new interface for every single API).

[ZAID] We are a server-side Mashup provider where users can extract  
raw HTML without the need for a remote Web Services API. They can do  
this programmatically or via our WYSIWYG tools. If they do it with our  
visual tools we automatically generate meta data.  That user- 
controlled meta data can be augmented, shared, and directly executed  
so every speech engine, desktop browser, widget/gadget system, and  
modern mobile such the iPhone is able to be driven by the meta data.

Our meta data does not need to be stored in an RDF database or queried  
by SPARQL for a user to derive knowledge.

>
>
> I'm a little confused by your talk about "alternate [commercial]  
> implementation of the SemWeb vision". Given that the Semantic Web is  
> a (data-oriented) extension of the existing (doc-oriented) Web, does  
> this mean you've made an alternate implementation of the Web  
> itself..? :-)

[ZAID] There is a difference between the SemWeb vision and its  
prescribed implementation which is codified by the various W3C SemWeb  
standards and literature. We've taken the updated TimBL vision-- to  
the best of our ability-- and implemented working code in the form of  
commercial software.

Today our users can generate meta data from just about every old  
broken HTML legacy web site that will never have RDF in it. And that  
meta data can be mashed up with other meta data and deployed in voice  
apps, desktop browser content, mobile apps, widget/gadgets...  
everywhere.

In conclusion,  today --with some exceptions and software bugs-- any  
user can generate meta data for the Billion + web pages for their own  
benefit or for the benefit of upstream SemWeb apps.


We just want to ensure that we can feed the higher level SemWeb apps  
with our Web 2.0 Mashup meta data... and that's why we came to you--  
hat in hand-- and said we changed a lot of stuff that you thought was  
irreplaceable but we've done some amazing stuff by going done an  
alternative path.

More truthfully though, doesn't just about every W3C SemWeb  
conversation end up with: "yeah, the SemWeb is going to dethrone  
Google with better ads and better search"? We preserve what's in the  
web today and don't require those billions of pages to be rewritten.  
So the W3C SemWeb prescribed implementation is more about changing WWW  
users and companies and not about supporting what's here today.


Our implementation empowers WWW users and also empowers the current  
Web 1.0 infrastructure companies such as the search providers, edge  
optimizers, web caches, and network routers. And a new breed of search  
companies can be spawned by providing element-based page analysis  
since as you saw in the Mashup W-Fi coffee example because we give you  
the exact data that's important for the user. Not just a document or  
reference meta data but the actual elements wrapped in meta data.

And no amount of machine intelligence can give you that.


>
>
> Anyhow I'll investigate further when I receive some advice on the  
> problem with the application.
>
> Cheers,
> Danny.
>
> -- 
> http://dannyayers.com
> ~
> http://blogs.talis.com/nodalities/this_weeks_semantic_web/



[ZAID] thanks for your patience and comments.


--Zaid

ALT Mobile

http://altmobile.com/Home.html (web site)
http://web.mac.com/altmobile (official blog)