On 7/26/12 4:29 AM, Michael Brunnbauer wrote:
Hello Kingsley,On Wed, Jul 25, 2012 at 05:27:21PM -0400, Kingsley Idehen wrote:Like I expected, she lost you in the second paragraph.Okay, and what exactly was that?Everything after "What is Linked Data?". Terms like "references", "structured data", "EAV". You may have lost sense of your altitude. Come down and draw a deep breath :-)
Okay, I can work on that part of the post. Note, I made a secondary post that just has a Turtle template.
Re. EAV, I used it with a broader audience than RDF in mind. What remains unbeknownst to many is the fact that EAV or EAV/CR is broadly understood and a lot more intuitive than RDF. Anyway, I'll see what I can do re. improvements etc..
But it would in fact be much harder because this time, she would have no reason to learn it and no tool to try out changes and see immediate *results*.Again, I beg to differ. If she has any desire to for her work to be discovered serendipitously without the markup tedium of HTML then Turtle is a potential short cut.How many people would actually use this data (via Sindice or some other means) ?
Good question.My answer: when you publish a document on the Web you don't necessarily do so with a single search engine / document indexer in mind. You just publish to the Web because you what to share something with others on the network, via the content of your document.
Ultimately, you have a cost vs benefits tradeoff re. making content comprised of simple statements using syntax like Turtle (which enables deployment anywhere) or the additional bloat of something like HTML into which you insert fine-grained structured data islands via RDFa and Microdata intermingling.
How many people use Google ? I know it's unfair but you cannot solve a chicken egg problem by declaring one of the two to be there.
As I said it isn't about Google. Its just about publishing documents because you want the content available to others on the network.
We have to be honest with ourselves about this technology. Whose problems does it solve ?I am going to give you a simple answer, one I am sure I've given in the past re. Linked Data. Problem: structured data representation, access, and integration across disparate data sources.Linear instead of exponential data integration cost is fine for big business and spooks but I don't think it plays a role in a SME. Show me business cases of RDF/SPARQL/OWL for SMEs.
The business case for Linked Data has always started by addressing the most basic business needs:
1. access to data across disparate data sources associated with line-of-business applications
2. conceptual level virtualization of disparate data sources via data objects that represent business domain specific and real-world entities -- customers, orders, competitors, employees, products, countries etc.. culled from a myriad of data sources (databases, soa services, network accessible documents etc..)
3. sophisticated integration at the conceptual level -- once you have a data virtualization layer of data objects you can deal with hard problems such as identity reconciliation and context fidelity matters like units of measurement, languages etc..
4. sophisticated data access policies -- data access rules can mirror enterprise policies and/or incorporate social dimensions e.g., constraining access to data based on enterprise hierarchy combined with social relationships (e.g., all executives that claim to know TimBL or vice versa can view this report etc.. )
5. effective dissemination -- every meme is beamed via a hyperlink (within and across the enterprise).
Of course, there's lots more.
Who can understand it ?Very simple, but that depends "who" is "whom" . Anyway, assuming the basic user of a computer that has saved a Word Document and an Excel Spreadsheet, you can explain as follows:Great explanation! But the way from understanding it to publishing, consuming or storing RDF is long and tedious.
We have to relegate the letters R-D-F to the background. This is about documents, content, and fidelity of content structure.
Are the tools usable in practise ?Of course they are. But I can't force you to accept that fundamental point even though there's been ample evidence of that since 2007.If you compare the tools with RDB technoglogy, they look bad.
You shouldn't compare them with RDBMS technology off the bat, we should demonstrate how they enable better use of RDBMS technology first. Once this beachead is established the rest of the value will manifest, without undue friction.
In the past, there's been a tendency to juxtapose RDF and RDBMS technology in manners the infer mutual exclusivity. Personally, I prefer to demonstrate how both technologies are mutually inclusive, because that's an eternal reality.
Let's get something done about it. How would you realize stateful SPARQL transactions via HTTP ?
What makes that definitive re. value proposition of Linked Data. At this point, SPARQL isn't important. The focus is establishing the fundamentals of Linked Data value propositions across a variety of scenarios. In this case, you are talking about the enterprise setting.
You may or may not realize, you are making the same mistake that you criticize. SPARQL doesn't need to be at the front door of this sort of conversation. At best, its an implementation detail, just like RDF etc..
Question for you, if you don't mind. Which of the following are useless, by virtue of questionable tools and/or utility comprehension? 1. World Wide Web 2. ODBC (Open Database Connectivity) 3. Java Database Connectivity (JDBC) 4. Microsoft Entity Frameworks (a component of ADO.NET) 5. X.500 Directories 6. Hyperlinks 7. URIs.Every one has it's use cases. Some have more, some have less.
But that doesn't answer my question. I wanted to know if any of the items above are useless, for instance. Basically, Linked Data compliments all of the above by improving their established utility.
The RDF/SPARQL/OWL stack has it's use cases and is here to stay.
But we are talking about usecases at that level. We are discussing the issue of Linked Data. Those items (once again) are *distracting* implementation details.
But we should not rise expectations that can't be met (now).
That's never been the case with Linked Data or Hyperlinks in general.
Regards, Michael Brunnbauer
-- 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
smime.p7s
Description: S/MIME Cryptographic Signature
