Re: "Microsoft Access" for RDF?

On 2/20/15 4:54 AM, Stian Soiland-Reyes wrote:
>
>
> On 19 Feb 2015 21:42, "Kingsley Idehen" <kidehen@openlinksw.com 
> <mailto:kidehen@openlinksw.com>> wrote:
>
> >> No, this is dangerous and is hiding the truth.
> > What?
>
> (Just to clarify my view, obviously you know this :) )
>
> That RDF Triples are not ordered in an RDF Graph.
>

Correct.

> They might be ordered in something else, but that is not part of the 
> RDF graph.
>

You can produce an order using:

select * where {graph <named-graph-iri> {?s ?p ?o}}
order by ?p

offset and limit can be used to create a paging mechanism, if required. 
Subqueries can be used for further optimize when the named graph is very 
large etc.

This UI would be for the user to alter relation subjects or objects. 
Predicate alternation isn't an option in this UI. Basically, the user it 
focused on relationship entities for a specific relationship type.

> (Reification statements can easily also become "something else")
>
UI leveraging Reification:

This provides an ability to let the user interact with a collection of 
statements in a UI where subject, predicate, and objects can be altered. 
Basically, they have a UX oriented towards sentence editing.

> So if you tell the user his information is just RDF, but neglect to 
> mention "and then some", he could wrongfully think that his list of 
> say "preferred president" has its order preserved in any exposed RDF.
>

Not the intent here at all.

> If you don't tell him it is RDF (this is now the trend of Linked Data 
> movement..), fine! It's just a technology - he doesn't need to know.
>

In our case we are showcasing RDF as Language, and using the UI/UX to 
bolster that point of view, using different UI/UX patterns to address 
the different ways a user can create or alter relations.

>
> > You can describe collections using RDF statements, I don't have any 
> idea how what I am talking about implies collection exclusion.
>
> My apologies, I got the impression there was a suggestion to control 
> ordering of triples without making any collection statements.
>

An RDF editor has to allow users create any kind of relation that's 
possible in the RDF Language.
>
>
> >> Don't let the user encode information he considers important in a 
> way that is not preserved semantically.
> > ??
>
> I simply meant to not store such information out of band, e.g. by 
> virtue of triple order or comments in a Turtle file, or by "magic" 
> extra bits in some database that don't transfer along to other 
> consumers of the produced RDF.
>

Okay, we don't do that. In fact, exposing relations as groups of rdf 
statements grouped by predicate enables clients lever optimistic 
concurrency patterns since they can make hash based checksums on the 
predicate based grouping that are tested prior to final persistence on 
the target store (SPARQL, WebDAV, LDP compliant).

> It should be fine to store "view"-metadata out of bands (e.g. which 
> field was last updated) - but if it has a conceptual meaning to the 
> user I think it should also have meaning in the RDF and the 
> vocabularies used.
>

We have a View that works with Controls en route to Data Persistence at 
storage location that supports one of: SPARQL 1.1 Insert, Update, 
Delete,  SPARQL Graph Protocol, LDP, and WebDAV . The Editor we've built 
is Javascript based. It also makes use of rdfstore.js, our generic I/O 
layer (also in Javascript) and a few other bits ontology lookups etc.., 
plus some bits JQuery integration etc..

> If you are able to transparently do the "right thing" semantically, 
> then hurray!
>

I think we do, but we'll see what everyone thinks once its released :)

>
> > Why do you think we've built an RDF editor without factoring in OWL?
>
> Many people are still allergic to OWL :-(
>

We aren't :)

> And also I am still eager to actually see what you are talking about 
> rather than guessing! :-)
>
>
> > I think we are better off waiting until we release our RDF Editor. 
> We actually built this on the request of a vary large customer. This 
> isn't a speculative endeavor. It's actually being used by said 
> organization as I type....
>
> Looking forward to have a go. Great that you will open source it!
>

Okay.

-- 
Regards,

Kingsley Idehen 
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog 1: http://kidehen.blogspot.com
Personal Weblog 2: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
Personal WebID: http://kingsley.idehen.net/dataspace/person/kidehen#this