On 11/14/11 2:16 PM, Len Yabloko wrote:
> On Mon, Nov 14, 2011 1:22 PM EST Kingsley Idehen wrote:
>
>> On 10/27/11 11:33 AM, Cory Casanave wrote:
>>> Thanks Peter,
>>> I have posted a suggestion on the ontology summit page as you suggested. I
>would also be happy to explore a tread on the topic and have therefor changed
>the title. The initial message, below, can serve as a problem statement.
>>>
>>> I would like to point out one clear fact: That with all the great work,
>tools, research and products available - the problem of information federation
>still exists and is getting worse. What we have now is either not working or
>not resonating. We don't need and probably can't produce a 100% solution - we
>don't have to. Making a 20% improvement in our ability to federate
>information and exchange data would be of immense benefit to companies,
>governments and society. I think we can do better than 20% and part of that
>is accepting that the 100% solutions are not currently practical. We have to
>make the solution set (of which ontologies are only a part), tractable and
>practical for widespread adoption - that has not been the track record so far.
>>>
> I think the core issue that is responsible for the slow progress in data
>federation is lack of reasoning capabilities supported by linked data
>standards. (01)
I think and hope you meant: lack of reasoning in a majority of Linked
Data platforms?
> This issue had Ben discussed many times on the ontolog forum. Ontologies
>differ from databases which offer sound and complete inference in feasible
>time/space. This why most common federation tecnnique is federated database. (02)
In my earlier post I purposely kept reasoning out, just to avoid
distraction. We use reasoning to handle data integration and data
quality issues. Simple example, handling labels for object when
different data sources use different properties/predicates inconsistently. (03)
Here's what I left out of my post;
*Basic SPARQL Update and Inference Rules based Tweaks*
In this case, we want data object identifiers (hyperlinks) to resolve to
pages where labels provide text for hyperlinks which increases user
friendliness and serendipitous discovery that results from
follow-your-nose style exploration patterns. (04)
Actual SPARQL commands:
-- add records (triples) to a named graph that will be used as the
target of an inference rule that deals with harmonization of object
labels, which is critical confining URIs to @href attributes of HTML
anchors when generating Linked Data oriented pages. (05)
INSERT into GRAPH <virtrdf-label>
{<http://demo.openlinksw.com/schemas/Northwind/categoryname>
rdfs:subPropertyOf virtrdf:label .
<http://demo.openlinksw.com/schemas/Northwind/name> rdfs:subPropertyOf
virtrdf:label .
<http://demo.openlinksw.com/schemas/Northwind/companyname>
rdfs:subPropertyOf virtrdf:label .
<http://demo.openlinksw.com/schemas/Northwind/lastname>
rdfs:subPropertyOf virtrdf:label .
<http://demo.openlinksw.com/schemas/Northwind/productname>
rdfs:subPropertyOf virtrdf:label .
} ; (06)
-- associate a named inference rule "facets" with the named graph
"facets" that holds the triple based rules outlined above. (07)
rdfs_rule_set ('facets', 'facets') ; (08)
The effect of the above is that the HTML based object descriptors
(presented as Web pages by Web Browsers) have URIs confined to @href
with the derived labels as anchor text. We pull this off on both the
TBox and ABox side of things, and its scales to billions of records [1]. (09)
The examples that follow are based on queries against our live 29
Billion+ triples LOD cloud cache instance. (010)
Links: (011)
1.
http://linkeddata.uriburner.com/describe/?uri=http://openlylocal.com/id/councils/167 (012)
-- example of a page where data has come from a variety of data sources
with reasoning used to harmonize labels (013)
2.
http://lod.openlinksw.com/describe/?url=http://linkedopencommerce.com/category/computers (014)
-- looking at information about computers (015)
3.
http://lod.openlinksw.com/describe/?url=http://www4.wiwiss.fu-berlin.de/dailymed/resource/dailymed/drugs&sid=105347 (016)
-- drug information from the Linked Open Drug Data project (017)
4.
http://lod.openlinksw.com/describe/?url=http://www.openlinksw.com/schemas/rdfs/
-- effect of linking DBpedia, Schema.org, SIOC, FOAF and other shared
vocabularies (018)
5. http://goo.gl/uuDyM -- an older G+ note about power of reasoning re.,
data integration . (019)
Kingsley
>
>>> This is a multi-billion dollar opportunity to address a pervasive and
>recognized problem. Let's get on with it.
>>>
>>> Regards,
>>> Cory Casanave
>>>
>>> -----Original Message-----
>>> From: peter.yim@xxxxxxxxx [mailto:peter.yim@xxxxxxxxx] On Behalf Of Peter
>Yim
>>> Sent: Wednesday, October 26, 2011 7:00 PM
>>> To: Cory Casanave
>>> Cc: steve.ray@xxxxxxxxxx; [ontolog-forum]
>>> Subject: [OT] process clarification [was - Re: [ontolog-forum] Some Grand
>Challenge proposal ironies]
>>>
>>> Cory,
>>>
>>>
>>>> [CoryC] An area of interest to me and many of our clients is solving the
>information federation problem. ...
>>> [ppy] A good topic indeed. However ...
>>>
>>> 1. if you are suggesting that folks discuss this "information federation
>problem" on [ontolog-forum], please consider starting a new thread (with a
>proper subject line) and move forward from there; or
>>>
>>> 2. if you are suggesting we (you addressing to Steve, following a remark of
>his regarding the Ontology Summit indicates that this might have been your
>purpose), it would be helpful if you condense the proposition to, say, a short
>theme/title, with a brief (short
>>> paragraph) description and post it to the
>http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit/Suggestions
>>> page (like what Christopher has done), and then, via a message post,
>highlight that suggestions, and take it forward similarly.
>>>
>>> (That would help allow this thread to stay on point to discuss what
>Christopher is trying here.)
>>>
>>>
>>> Thanks& regards. =ppy
>> Cory,
>>
>> I've just published a note on G+ about enterprise Linked Data and how it
>> addresses the thorny issue of data virtualization across heterogenous
>> data silos. As per usual, I've included links to live examples that
>> showcase the effects outlined in the post.
>>
>> We have to be able to loosely couple data, information, and knowledge.
>> This is something hyperlinks handle very well once we get past
>> distracting syntax wars re. how to represent data objects via directed
>> graph pictorials. etc..
>>
>> Links:
>>
>> 1. http://goo.gl/8I3KB -- G+ post about the importance of enterprise
>> linked data .
>>
>> --
>>
>> Regards,
>>
>> Kingsley Idehen
>> President& 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
>>
>>
>>
>>
>>
>>
>
> _________________________________________________________________
> 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
>
> (020)
-- (021)
Regards, (022)
Kingsley Idehen
President& 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 (023)
smime.p7s
Description: S/MIME Cryptographic Signature
_________________________________________________________________
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 (01)
|