Re: 15 Ways to Think About Data Quality (Just for a Start)

On 4/13/11 10:35 AM, glenn mcdonald wrote:
>
>     Using Safari, when I click on the link above I hit an
>     authentication challenge [1].
>
>
> Oh, I see. Clicking on the export links in the browser works without 
> authentication, but issuing the queries directly doesn't yet. No good 
> reason for that, so I'll get it changed.
>
>     1. Do you have a Data Object Identifier (Name Reference) for
>     'Michael Jackson' that's distinct from its actual Object
>     Representation Address (Data Location Reference) with regards to
>     the entity 'Michael Jackson' ?
>
>
> Yes. I already answered this twice already.

Okay, so maybe I just don't understand your response. If you could, 
please respond with links for the following:

1. Object ID
2. Object Representation Access Address.

I would like to click (or HTTP GET) on #1 and arrive receive a 
description graph courtesy of name-address indirection that occurs 
between #1 & #2.

>
>     2. Does the fidelity of the Object Graph that constitutes the
>     Object Representation remain consistent across data representation
>     formats?
>
>
> Mostly. There's a "Plain Text" format that doesn't include any 
> relationships, and the other formats differ slightly in how they 
> represent one-to-many relationships, but the logical structures are 
> the same.

Does the Object ID, once de-referenced, deliver to me a graph that 
describes the Referent (in this case 'Michael Jackson') ? Basically, can 
I discern the nuances you outline above via the object representation graph?

>
>     3. Can I HTTP get against the Object Identifier for 'Michael
>     Jackson' and receive a graph that describes 'Michael Jackson'?
>
>
> Yes. You'll probably have better results asking for the information 
> you want, rather than relying on the default response, but that's a 
> different question.

For purposes of me trying to align our world views, the default response 
has to deliver me an object representation graph. No problem if it 
doesn't, but note it will prove our conversation still suffers from a 
serious "Apples vs Oranges" problem.

>     The questions above are important since they define the HTTP based
>     Linked Data concept and also lie at the core of all URLs published
>     re. my demos. This is how we get to a more "Apples vs Apples"
>     style of conversation.
>
>
> But to be clear, this is not my goal.

Yes, but be clear, this is my goal. It defines the essence of every 
single demonstration I publish with regards to Linked Data. Remember, 
100% of the time you've responded to my Linked Data demo links.

> Needle makes no claims to be a Linked Data brand system.

I never said it made such a claim. What I've told you repeatedly is this:

1. My demos are about showcasing Linked Data prowess
2. Linked Data prowess is about the power of Links
3. Links are a powerful mechanism for conducting Data
4. Data conduction is distinct from Information conduction
5. Data conduction is distinct from Knowledge conduction.

1-5 define the essential elements of my subjective context halo.

> It is HTTP/REST-based, it is a graph database, and every node does 
> have an ID (and by virtue of this a URI), but linking is not our 
> primary concern.

Again, Linking via use of Links for "whole data representation" is my 
primary concern. It's also what Linked Data is all about.

> You're actually the one who brought Needle into this conversation, not 
> me.

No, I am trying to get to an important point where one of the following 
materializes with clarity:

1. Prove to you we are still having an "Apples vs Oranges" issue re. our 
conversation
2. Prove to you we are having an "Apples vs Apples" conversation.

It's ultimately one or the other.


> Ironic, given your short temper about people derailing "your" threads.

What can I say to that? "You" continue to tell "me" about "myself".


-- 

Regards,

Kingsley Idehen 
President&  CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen