The New Semantic Web ?


Recently there has been some renewed interest in the Semantic Web.  This new interest, me thinks, has been fueled in part by Tim Berners-Lee's recent article about Putting Government Data online, which in turn, me thinks, was inspired by  Obama's pledge to introduce more transparency into government.  I, of course applaud this new interest and hope that, this time it will help us realize last decade's dreams of a web of data as useful as the WWW.

But I wish to point out, again, the major flaw that i have identified in the Semantic Web (SW).  Me thinks that, if this flaw is not fixed, those dreams cannot be materialized. The flaw is that the triple (subject, verb, object) is insufficient to represent true facts.  The triples are the building blocks from which the data and facts of the Semantic Web are built.  They are the very stuff of that web, just as documents containing hyperlinks to other documents are the stuff of the WWW.  But if true facts cannot be contained in the triples of the Semantic Web, how then can we expect our government to make its truths available to us in that web of data?

Why, you may ask, is a triple insufficient for representing true facts?  Well a statement of fact is true within the context of an individual or group who believes or asserts it.  The subject-verb-object (svo) of the triple is quite adequate to assert almost any fact that can be represented in language.  But that triple is insufficient to assert the truth of the triple; for that we need to identify the context of individual or group who believes it or is asserting it.  Consequently, I, and others, have, from time to time, proposed putting an additional element identifying the context (or ownership) of the triple making it, in fact, a quad

Outside of the epistemological reasons for the triple's insufficiency there is another more urgent and practical reason why it would help the progress of the Semantic Web for it to update the triple to a quad.  That reason is ownership and responsibility of data.  In this wild west of the Web, where anyone can and will say anything about anything, a simple triple just cannot stand on its own.  Yet that is the basis of the current Semantic Web.  You should be able to convince yourself of this by examining a couple of triples yourself.  May i suggest: (Bush is bad) or perhaps (Obama is good).  Can there ever be any truth is those triples?  No, I don't think so.  Now let me convert those to quads that you should immediately recognize as true: (Seth-says Bush is bad) and (Seth-says Obama is good).  Now can you doubt the truth of those statements? 


To be continued ...


Tags

  1. semantic web
  2. open government
  3. tim berners lee
  4. rdfa
  5. linkeddata

Comments


Mark de LA says
Your 2 red herring statements don't exemplify any more truth in the quads than the triples. You would have to build a whole nest of quad sticky molecules which answer who is Seth etc. to get to any truth & then it might all dissolve into bs. Context is a distinction of it's own needing triples & quads to define.
Putting data online is good if you have the power to support mass inquiries. When the recent cap & trade bill was put on line the website was inaccessible. We need something better than text search, but only an elite few will understand quads & SVO. More work is needed at the front (user) end. Teradata over 20 years ago had a system to guarantee a response time of ~ 60 sec per inquiry no matter how massive the database was. I'm sure they & others have improved that since then. The level of intelligence in the population is at the google keyword search - what next?


Seth says
A lot of the new discussions about the SM seem to involve the new RDFa which appears to be a method of embedding RDF triples inside of XHTML. 

Hmmm, i wonder if one can also embed a context identifyer in those XHTML documents such that quads instead of triples could be extracted.

Mark de LA says
I think that a web-based CyberMind with a cleaner, easier CyEnglish MPL would fit Tim Berners-Lee's fascination with links. That's still one of my 101 ideas of things to do in my group projects when I get .
The language is old and goofy although gress, vance, sym, mem & remem are stable but could use updating to web context & at to URI. I wonder if there is something out of text messaging slang that is useful here.

Seth says
MR 2009-07-08 06:33:07 12210
Your 2 red herring statements don't exemplify any more truth in the quads than the triples.
Mark you really should try to understand what i am saying instead of just arguing about it.  This is not a partisan debate, rather it is about linguistics.   Let me give you a different example that may bypass some of your preconceptions about what i am saying. 

If i say "My cat Fluffy is on the mat" that statement is only true just so long as Fluffy stays on the mat.  When she dashes off to go pee, the statement suddenly looses verity.   But if I say "My cat Fluffy is on the Mat at 8:35 AM Wednesday July 8 2009", then when she goes to take a pee, the statement is still true.  In fact, if the statement was ever true, it will stay true forever. That linguistic fact was not invented by me but was described ad nauseam by the likes of Bertrand Russell and Willard Quine in the early part of the last century.

All i am saying here is that by adding context (ownership, responsibility) to statements we can improve their application.  We then would have official containers into which we can add and update statements.  The statements don't have to float in the cloud of the Semantic Web purporting to be universally true wherever we find them.  No, they are only deemed true to the authors of the containers.  Facts will become the responsibility of those people who have the passwords to write and erase in their official containers. 

My point is actually more practical and technical than philosophic and political.  When (as and if) we start to navigate the New Semantic Web of government data perhaps i can come up with more down to Earth examples ... perhaps even where things are going bump in the night.  From a the point of view of users navigating data on the SW, they should always be able to navigate easily to who is responsible for that data. 

Seth says
MR 2009-07-08 09:55:46 12210
MR 2009-07-08 09:31:02 12210
Yep, thanks for the better example. You still miss the point that a quad or a pent or any vector of data is insufficient without much more of a cluster to accompany it. A whole natural language sentence is rarely sufficient to pin down a truth. A cat is on the mat at a time is unintersting with regard to where the mat is etc.
Back when I was inventing an event database I noticed that the question words: who, what, when, where, why & how were a good start but even more question words exist.

Well certainly i agree that the more information you have about a subject the closer you will be to what it is all about.  That really goes without saying. 

Back to my point, most of the data on the SM will probably be of the nature (TTHERMO price $169.95) which doesn't make much sense taken as a triple.  But if you can see that in the context of my feed http://www.speaktomecatalog.com/feed/speakatom.xml it makes a whole lot of sense.  In other words, a factoid gains its meaning from the container in which you find it.

Seth says
Re: yahoo search using RDFa labeling.
source: My post to the W3C mailing list
Well at first blush this is good news.  But then why do we want to use
a  vocabulary like product:listPrice which is locked to a specific
yahoo domain: http://search.yahoo.com/searchmonkey/product/ ?  Is
Google going to come along with another different word for that?  Then
which one does a catalog owner choose?

Re this example from the referenced page:

<div typeof="product:Product"
   xmlns:product="http://search.yahoo.com/searchmonkey/product/"
   xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#">

   <span property="product:listPrice">
27.99</span>
   <span property="product:currency" content="USD" />

   <span property="rdfs:label">Startech Serial ATA Cable - 45.72cm - Red</span>
</div>

Fact is i was laying in bed this morning thinking that it might be a good idea for me to add these kind of RDFa tags to my catalog.  It wouldn't take all that very much since it is all automatically generated from the db.  Thing is now which vocabulary should i use?


Seth says
Martin Hepp has a new example of marking up catalogs with the GoodRelations vocabulary. Here is my response on the group ...
source: my email to the semantic web mailing list
Thanks for the example :)  It looks quite doable.

Do you know if someone has written a validator page for this RDFa kind
of markup?  We need something like http://validator.w3.org/feed/ .
There is nothing like instant feedback ... a page that you can go to
and paste in your URL and get back a web version of the triples that
it creates as well as any errors.

--
Seth Russell
www.speaktomecatalog.com

...


Seth says
Tim Berners-Lee puts the whole #linkeddata  (Semantic Web)  craze in relationship to Open Government in this youtube video Open, Linked Data for a Global Community

See Also

  1. Thought Browsing the Semantic Web with 77 viewings related by tag "linkeddata".
  2. Thought Sourcing Around with 35 viewings related by tag "open government".
  3. Thought JasonLd with 14 viewings related by tag "SemanticWeb".
  4. Thought about: dbpedia live, via openlink virtuoso, in the amazon web services cloud with 11 viewings related by tag "semantic web".
  5. Thought What is the mentography of JSON ? with 6 viewings related by tag "linkeddata".
  6. Thought Privacy is antithetical to the semantic web with 5 viewings related by tag "semantic web".
  7. Thought Quads with 3 viewings related by tag "semantic web".
  8. Thought Naming Context with 3 viewings related by tag "linkeddata".
  9. Thought Beyond News Fronts with 2 viewings related by tag "semantic web".
  10. Thought The Connections & Relationships Project with 2 viewings related by tag "semantic web".
  11. Thought about: MindRaider - Semantic Web Outliner with 1 viewings related by tag "semantic web".
  12. Thought about: google open source blog: cayley: graphs in go with 1 viewings related by tag "linkeddata".
  13. Thought funny page cartoon with 1 viewings related by tag "semantic web".
  14. Thought about: paper.li - read Twitter as a daily newspaper with 1 viewings related by tag "linkeddata".
  15. Thought about: Re: How will the semantic web emerge from Seth Russell on 2005-12-13 (semantic-web@w3.org from December 2005) with 1 viewings related by tag "semantic web".
  16. Thought about: SemNews - Semantic News Framework with 1 viewings related by tag "semantic web".
  17. Thought Tagging Is A Better Memory Substitute Than Hyperlinking with 0 viewings related by tag "tim berners lee".
  18. Thought Searching & Writing for the Semantic Web with 0 viewings related by tag "semantic web".
  19. Thought About Blogging & Wikis from Sir Tim Berners-Lee with 0 viewings related by tag "tim berners lee".
  20. Thought Social Media Space Agnostic Comments Demo with 0 viewings related by tag "linkeddata".
  21. Thought From the News with 0 viewings related by tag "tim berners lee".
  22. Thought about: Identity, Reference, and the Web (IRW2006) Workshop with 0 viewings related by tag "semantic web".
  23. Thought about: Change.gov with 0 viewings related by tag "open government".
  24. Thought How is white house data processing going to change? with 0 viewings related by tag "open government".
  25. Thought about: me at sig.ma with 0 viewings related by tag "linkeddata".
  26. Thought about: data on the web best practices with 0 viewings related by tag "semantic web".
  27. Thought about: linked open vocabularies with 0 viewings related by tag "semantic web".
  28. Thought Semantic Web? Questions with 0 viewings related by tag "semantic web".
  29. Thought Commercial aspects of Smart Agents and Bots with 0 viewings related by tag "linkeddata".
  30. Thought Why I left the Semantic Web! with 0 viewings related by tag "semantic web".
  31. Thought Me represented in the Linked Data Cloud with 0 viewings related by tag "linkeddata".
  32. Thought about: usdot announces new, historic uniform high-speed rail design standards to help u.s. manufacturers to compete with 0 viewings related by tag "open government".
  33. Thought UI for client cert selection (Was: Releasing RWW.IO) with 0 viewings related by tag "tim berners lee".
  34. Thought about: watchdog.net with 0 viewings related by tag "semantic web".
  35. Thought Hyperfast Analysis of Legislation (HAL) with 0 viewings related by tag "rdfa".
  36. Thought about: timbl's blog | Decentralized Information Group (DIG) Breadcrumbs with 0 viewings related by tag "tim berners lee".
  37. Thought Serendipituous playing in open LinkedData Spaces with 0 viewings related by tag "linkeddata".