I admit to not following this thread closely, so I apologize if this is
off-topic or repetitive. I further admit that I abandoned (or--as I
thought at the time--deferred) the use of formal ontologies to help
manage information in my space (PLM/ERP) a couple of years ago. (01)
Nevertheless I am keenly interested in the problem of data integration
in an RDF ecosystem. For many practical applications, but especially
those that eschew formal class definitions, what matters is not
equivalence of (formally defined) classes, but rather equivalence of
statements (or, by extension, groups of statements in an RDF graph).
This is more like "data-driven data integration" than ontology-driven. (02)
You want to be able to say that the class of RDF statements such that
one set of property restrictions holds is equivalent to the class of RDF
statements where some other set of property restrictions holds. Or, more
generally, you want to assert such equivalence of graphs. (03)
I first encountered this problem after merging product data from the PLM
system, the shop-floor manufacturing system, and the ERP system. To take
one example: Each system has a different property for "unit of measure",
and a different range of values for the property. Say the properties are
plm:UnitOfMeasure, ms:UOM, and erp:MEINS (can you guess what the ERP
system is?). Congruent values in the respective systems might be "EA
(each)", "EA", "EA"; "OZ (ounce)", "OZ", "OZ"; "FT^2 (square foot)",
"SQFT", "FT2". (04)
So in the data I might find:
<http://example.org/plm/part/12345> plm:UnitOfMeasure "EA (each)" .
<http://example.org/ms/part-master/12345> ms:UOM "EA" .
<http://example.org/erp/material/12345> erp:MEINS "OZ" . (05)
I want to be able to detect cases like the 3rd statement, which is not
in accord with the other two. (06)
My preferred approach now is to use rules, but I found you can
accomplish nearly the same thing in RDF by using the reification
vocabulary. (07)
First you add assertions to the repository like the following. (08)
[a rdf:Statement;
rdf:predicate plm:UnitOfMeasure;
rdf:object "EA (each)"]
ex:equivalentPredicate [a rdf:Statement;
rdf:predicate erp:MEINS;
rdf:object "EA"] . (09)
Insert similar statements for each pair of equivalent predicates and
values. (My OWL is too rusty at the moment to attempt to define the
conditions using owl:equivalentClass.) (010)
If I have some way of asserting that the subjects all denote the same
thing (which is easy enough with SKOS), then a sparql query like this
reveals the discrepancies: (011)
select ?entity1 ?uom1 ?entity2 ?uom2
where {
?entity1 skos-xl:prefLabel ?xlabel;
plm:UnitOfMeasure ?uom1 .
?entity2 skos-xl:prefLabel ?xlabel;
erp:MEINS ?uom2 .
filter not exists {
[rdf:predicate plm:UnitOfMeasure;
rdf:object ?uom1] ex:equivalentPredicate
[rdf:predicate erp:MEINS;
rdf:object ?uom2] .
}
} (012)
In this case I chose to use skos-xl:prefLabel to indicate that the
subjects "are about the same thing". There are other ways to do this,
either more or less formally. (013)
As noted above, the same results can be achieved using rules, and in
fact in my work I rely more on rules than on notions of equivalency. But
since the problem statement said "using owl:equivalentClass" I thought
this approach might be relevant. But I have only done it in RDF; there
may be unforeseen difficulties putting it into OWL. (014)
Regards,
--Paul (015)
On Sun, 2014-02-09 at 08:40 +0200, Uri Shani wrote:
> Hello Kingsley and friends. I'd like to open here a clean page in
> reference of your suggested solution in
> http://ontolog.cim3.net/forum/ontology-summit/2014-02/msg00117.html#nid09.
> I do that since the mailing chain is pretty hard to follow by
> now.... :)
>
> The problem example you state is to reason about the concept
> 'Organization' over two different (but related) ontologies A and B
> where a:Ogranization and b:Organization represent close and
> possibly semantically equivalent concepts.
> Your solution is to define "integrating ontology" C which imports the
> two ontologies A and B, and defines the triple [a:Organization
> owl:equivalentClass b:Organization .], which will now
> allow to reason with C about this concept as an integration over A and
> B.
>
> Objections to that approach have been that there will be conflicts
> among A and B that will entangle that reasoning in C and possibly fail
> it or produce mistakes. I would add to this that
> this will also add much overhead and inefficiently slow process.
>
> Alternative solution suggestion:
> Lets define a common ontology O in which we have a single concept
> o:Organization. Now we define "integrating" ontologies AO and BO to
> bridge A and B with O:
> AO imports both A and O, and defines the triple [a:Organization
> owl:equivalentClass o:Organization .]
> BO imports both B and O, and defines the triple [b:Organization
> owl:equivalentClass o:Organization .]
>
> The bridges work as two separate mediators between A and B to a
> "common" ontology in which all organizations in A and in B will be a
> o:Organization rather than a a:Organization or a a:Organization,
> respectively.
> How is this useful? It depends on your application:
> We will build two mediation "networks": one flowing models of A to O1
> (a model in ontology O), using the integrating ontology AO, and
> another from B to O2 (another model in ontology O), using BO. These
> are like views of the original models.
> At this point one can work on a federated data-set over the two models
> O1 and O2, being of the same ontology.
>
> That is an analogy to what happens in DBMS where schema play the role
> of ontologies.
>
> The power of the mediation process is in more elaborate examples,
> where O may be covering many more important concepts than just
> Organization, and represent a meaningful knowledge about organizations
> that overlaps A and B (and possibly many more ontologies in this
> domain). Models in O can be transformed (mediated) from all these
> ontologies, and what is interesting here: also mediated back to their
> original ontologies. The "combining" ontology is in our terminology -
> a "Mediation Rule" ontology.
>
> About our problem domain:
> We deal with interoperability among engineering design tools, where we
> wish to share models in one tool with models in another tool, with no
> limit of the umber of tools. So we defined a hierarchy of commonality
> among ontologies and with the mediation rules ontologies we are able
> to mediate models between specific tools and models in the common
> ontology - back and forth.
> For instance, a model in the SysML language is mediated to a tool
> implementing the Modelica equations modeling method - and back
> creating a collaborative process among engineers working on these two
> very different tools on the same engineering design.
>
> So, I suggest to apply the same approach to the problem discussed in
> this mail thread.
> For more, see our paper: H. Broodney, U. Shani, A. Sela, “Model
> Integration – Extracting Value from MBSE,” in MBSE, INCOSE
> International Symposium 2013, Philadelphia, U.S (best paper):
>
>https://www.researchgate.net/publication/255720194_Model_Integration__Extracting_Value_from_MBSE
>
>
>
> Regards,
>
> Uri Shani, PhD
> Research Staff Member
> SPRINT(lead), DANSE (lead) Projects
>
> ___________________________________
> Phone:972-4-829-6282 |
> Phone:972-4-8296282 |
> Mobile:972-54-697-6282
> E-mail: SHANI@xxxxxxxxxx
> Find me on: LinkedIn:
> http://www.linkedin.com/in/urishani Facebook:
>http://www.facebook.com/uri.shani
> IBM
>
> Haifa University, Mount Carmel
> Haifa, HA 31905
> Israel
>
> _________________________________________________________________
> Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
> Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
> Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
> Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
> Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014
> Community Portal: http://ontolog.cim3.net/wiki/ (016)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014
Community Portal: http://ontolog.cim3.net/wiki/ (017)
|