ontolog-forum
[Top] [All Lists]

[ontolog-forum] Using Ontologies for Legacy Data Virtualization [was Sel

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Len Yabloko" <lenya@xxxxxxxxxxxxx>
Date: Sun, 15 Apr 2012 20:46:15 +0000
Message-id: <W195941110767841334522775@webmail37>
Hi Everyone. I took a liberty to spawn a new thread.     (01)

This is a topic of great interest to me and also the area where I have some 
expertise. Allow me to share some thoughts on what had been discussed below.        (02)


>On 4/8/12 1:41 PM, David Eddy wrote:
>> Kingsley -
>>
>> On Apr 8, 2012, at 12:30 PM, Kingsley Idehen wrote:
>>
>>>> I am assuming:
>>>>
>>>> (1) you (the linker) do not know
>>>>
>>>> (2) you do not have access to the person who does know
>>>>
>>>> (3) the person who does know is likely only partially correct.
>>>>    (03)

These assumptions do not even need to be made explicit since there are arguably 
already implicit in any discussion about data. Besides, (3) is all you need. It 
had been said and repeated many times that "all models are wrong, but some are 
useful". It had been also said way before that "meaning is use".        (04)

>>>>
>>>>
>>>> SemWeb is looking for a presentation Tuesday.  Can you show how 
>>>> Linked Data resolves this "semantic connectivity" issue?
>>>
>>>> I just sent a reply to Rich, I forgot to make it clear that I was 
>>>> also responding to you. If my posts don't clarity matters, I'll 
>>>> address the questions you posed further.
>>
>> Making the actual---physical---connection is of course crucial. 
>>  Having the pipe between point A (point of origin) & point B 
>> (destination) is fundamental to the data connectivity challenge.
>>
>> It is certainly interesting (that's not a positive) that ODBC was 
>> limited to RDBMs data sinks.  Definitely kudos for expanding the ODBC 
>> connectivity universe.
>>
>>
>> Let me offer a smidge of context... if I'm talking to a technical/data 
>> person at a Fortune 500 shop & they tell me they're a Database X shop, 
>> I instantly know I'm not talking to the right person.  The correct 
>> answer is "Yes."  We have it---DB2, IDMS, M204, S2000, IMS, Oracle, 
>> Sybase & of course the Mother of all databases the $2M Excel 
>> spreadsheet---pretty much anything & everything.
>>
>> I ran into a guy who confessed his shop had 28 DBMS engines... I can 
>> only get to about 17... ignoring desktop, but widely deployed, engines 
>> such as SQL Server, Rbase, Mimer, 4D, Filemaker, Postgres, XBase, 
>> etc., etc., etc.
>>
>> Folks who assume the corporate world is "relational" clearly don't get 
>> out enough.  IBM's IMS product still generates US$700M in revenue. 
>>  From the customer's perspective, it's bought, paid for, & performs 
>> within know boundaries... why endure a 10 year migration to another DBMS?
>>
>> This background is only to point out that legacy silos are here to 
>> stay... eventually they do fade away & get unplugged, but far more 
>> often additional layers are wrapped around existing silos.
>>
>>
>> So the SEMANTIC CONNECTIVITY challenge becomes... [drum roll...]  How 
>> do I know what I'm putting into the pipe at Point A when I want to tap 
>> a legacy source?  See the above 3 questions.
>
>You virtualize the heterogeneous data sources such that all the records 
>in said data sources become data objects. Then you leverage the EAV 
>model such that each data object is endowed with the following:
>    (05)

What exactly is "virtualize"?. I believe this term has no well-defined or even 
commonsense use. The only use of that term I am aware is for marketing of 
one-size-fits-all solutions to under-defined challenges.      (06)

>1. Unique Identity    (07)

What do you mean by Unique Identity? Are you suggesting that any of existing 
technologies can guaranty that different references are not pointing to same 
entity, or that one reference points to only one entity. That is only possible 
if you took special care of "closing" what otherwise is an "open world 
assumption" required outside individual data silos. Such technologies are still 
in research stage and none is implemented, as far as I know.       (08)

>2. Structured Representation -- an EAV graph pictorial    (09)

What part is provided by Linked Data - graph of pictorial? The graph is simply 
a data model deemed useful by publisher of data. It is not provided by some 
technology but merely exposed or converted in some format. No semantic 
processing is done here. As for pictorial, it can be rendered from any 
structured data liked or not.     (010)

>3. A Variety of Representation Syntaxes    (011)

There no value added here beyond well-developed ETL technologies    (012)

>4. Access Address    (013)

Access to what? That is the question. It was in fact the question asked at the 
top (or bottom) if this thread    (014)

>5. Negotiable Data Serialization Formats -- i.e., in response to an HTTP 
>GET.    (015)

That is the part of HTTP, not part of any derived technology.    (016)

>
>Apply HTTP scheme URIs to #1 and #4 and you enter the RDF model realm.
>    (017)

And how is this "realm"(another useless word) helping?    (018)

>Mandate that the data object URIs above resolve to data object access 
>addresses and you have Linked Data.
>    (019)

But why do you need to have it? How can it be used?    (020)

>>
>> And there should likely be a #4... the data source is likely not 
>> webified/HTTP enabled.
>Exploit R2RML [1], and you have a declarative syntax for mapping RDBMS 
>data sources to Linked Data.
>    (021)

This does no enhance any usefulness of Linked Data. It makes it easier to 
produce something that is not very useful.    (022)

>Leverage what we do with cartridges/drivers [2] for transforming other 
>data sources to this EAV + URIs model and you have Linked Data extending 
>beyond the RDBMS.    (023)

"Cartridges"? Is that something you need for "virtualization"?     (024)

>
>Links:
>
>1. http://www.w3.org/TR/r2rml/ -- R2RML mapping language
>2. http://virtuoso.openlinksw.com/dataspace/dav/wiki/Main/VirtSponger -- 
>Virtuoso Sponger .
>
>Kingsley
>>
>> ___________________
>> David Eddy
>> deddy@xxxxxxxxxxxxx <mailto:deddy@xxxxxxxxxxxxx>
>>
>>
>>
>>
>>
>>
>> _________________________________________________________________
>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>> Shared Files: http://ontolog.cim3.net/file/
>> Community Wiki: http://ontolog.cim3.net/wiki/
>> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>>   
>
>
>-- 
>
>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
>
>
>
>
>
>    (025)



_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (026)

<Prev in Thread] Current Thread [Next in Thread>