Re: Using "Punning" to Answer httpRange-14

On 5/15/12 4:19 AM, Michiel de Jong wrote:
> Hi Kingsley,
>
> i would love to learn more from what you wrote, but as a non-native
> speaker, and on top of that a newcomer to the world of web
> architecture, i have quite a bit of trouble understanding what you
> mean. let me point in your message where i did not understand what you
> are saying.
>
> On Mon, May 14, 2012 at 3:12 PM, Kingsley Idehen<kidehen@openlinksw.com>  wrote:
>> There are two types of 303 people:
>>
>> 1. Those that wander into the realm
> which realm? wander in which sense?

I meant: wonder . Existing Web users wonder into Linked Data spaces on 
the Web. For instance, existing Web users work fine with DBpedia 
irrespective of their comprehension of Linked Data principles.
>
>> 2. Those that have opted to support non # URIs using this pattern.
> ok, so from this i understand the hash-uri-rule is a default from
> which 303 diverges? interesting, i had considered them both candidate
> proposals, each with their own limited following. so i think what
> you're saying with 'support non # URIs' is:
> - if the URI contains a #, then it's always to be interpreted as
> object-sense, never as object-content.

Sorta, but in reality there is just an object-sense i.e., object name. 
When dealing with Data we only care about Names. It is important for 
people to understand that data objects have names that resolve to 
content. The content in question (re. Linked Data) takes the form of a 
machine/human discernible EAV/RDF graph.

Anyway, to answer your question specifically, hash (#) URIs are a low 
cost generic naming mechanism baked into AWWW (Architecture of the World 
Wide Web). Low cost because of *implicit* indirection re. name 
de-reference since # never cross the wire.

> - if it doesn't then the default (hash-uri-rule) is to interpret it as
> object-content, *but* 303 people of type 2 will still interpret
> object-sense if retrieving the URL results in a 303.

Your object-content == what I prefer to call a Web resource .  When an 
HTTP  server returns 200 OK it means you are dealing with a Web (realm) 
resource i.e,. a Web document. These too are *things* that can have a 
name. It just so happens that those names end with a "/"  (slash) .



>
>> #1 could care less
> i'm familiar with the expression "couldn't care less", but care about
> what? '#1' refers here to the 303 people who wonder into the realm i
> think, but are you saying the do care or they don't care?

Most users see hyperlinks as Web Resource (document) names. The don't 
care about anything other detail.
>   and what is
> it they care about or not? i am reading this for the second time now
> and it's really too abstract for me to parse, please explain this to
> me as if you're talking to a 6 year old.
>
>> since they haven't even bought into
> see below
>
>> the concept of semantic fidelity
> i searched for that concept and saw it used in some places but could
> not find any definition of it. do you have a link? i would love to
> learn! i'm assuming it means to make it less ambiguous (to a machine
> and/or to a human) what a certain document means.

relations and their semantics. RDFS provides low semantic fidelity for 
property descriptions. OWL provides high semantic fidelity. Substitute 
"fidelity" for "granularity" .

>
>> via structured content bearing
> i'm assuming the structured content here could be like a web page,
> right? so we're talking about adding

A Web page is the product of structured content (HTML) delivered from an 
network address via a server in response to a client request.
>
>> relational property graphs.
> to web pages. i searched the web for 'definition "relational property
> graph" ' and found this link
> https://plus.google.com/112399767740508618350/posts/LwEb8ccs8on where
> the term is used (by you actually! :) but i could not find any clear
> definition. so for the 3rd time within one sentence i find myself
> guessing at the definition of the phrases you use. my best guess would
> be that a relational property graph is a graph that expresses how
> objects relate to each other, and you label outgoing edges as
> properties of that node, like the "parent" property of node A is a
> pointer to node B iff node B is the parent of node A. So i guess
> adding such information to structured content is like what we do if we
> add a 'property' attribute to a link or to an html element in a web
> page, right?

Nice place to lookup relatonal property graphs in relation to DBMS tech: 
http://www.slideshare.net/slidarko/an-overview-of-data-management-paradigms-relational-document-and-graph-3880059/26 
.
>
> ok, so putting that all together, what you're saying is that the
> people who view 303 as a fallback for hash-uri-rule have not bought
> into adding semantic markup into the document itself.

I am not saying that per se. I am saying that we have object names and 
it just so happens that these names need to be realm agnostic i.e., you 
have objects on the Web and you have objects in the real world. When 
dealing with the Web resources solely, name/address ambiguity of HTTP 
URIs doesn't matter since an object address can also be its name. When 
dealing with entities or things not of the Web then we hit ambiguity 
when using "/" HTTP URIs, so there is a need for explicit disambiguation 
on the part of the Web resource publisher. Note, the Web resource 
publisher is the producer of Web resources that bear content that 
explicitly describe some unambiguously named real world or web entity 
via structured content in EAV/RDF graph pictorial form.
>
> that's confusing to me. why would you even care about if a link is
> object-sense or object-content if you don't know the "flavour" of the
> link?

"I" the publisher of resources to the Web where my clients are Linked 
Data oriented should care about the object names I publish.
"I" the developer of a Linked Data client has no choice but care about 
when a URI is a Name or an Address otherwise my Linked Data exploitation 
is impeded (obstructed).
"I" the Web user doesn't care about any of the above until I realize 
that I am at a disadvantage to others re. my ability to make sense of 
data on the Web.
>
> i did try to read the rest of your post, but i didn't understand it
> either. i hope this reply makes clear that i really don't understand a
> word of what you're saying! :) i stumble and have to make assumptions
> about what i think you might mean, several times per sentence.
>
> can you maybe summarize your post in language as if you are talking to
> a 6-year-old non-expert who is at the same time not a native speaker?
> of course i'm exaggerating, but it's really true that i can't learn
> from you if you speak in the language you speak in.

A picture speaks a thousand words. I published an illustration yesterday 
[1]. You can fork alter and share. This matter is sometimes best ironed 
out via illustration.

The trouble with this whole subject matter realms is that its vast. It 
back to front because many don't seem to understand that the Web is an 
application of existing computer science rather that a new subject 
matter realm on to itself. This is all about data objects and networks.

Links:

1. 
https://docs.google.com/drawings/d/1ZUzBa4HjNUXg_OeFudwK0XO70VeJRxJoXv4RW2KamhY/edit  
-- illustration of what happens with names and indirection re. Linked Data

2. 
http://www.slideshare.net/slidarko/an-overview-of-data-management-paradigms-relational-document-and-graph-3880059/26 
-- relational property graphs

3. http://lists.w3.org/Archives/Public/public-lod/2012Apr/0081.html -- 
old post about Data and Networks which has links to a presentation that 
covers the crux of the matter well .

Kingsley
>
> cheers!
> Michiel
>
>
>


-- 

Regards,

Kingsley Idehen	
Founder&  CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen