search

spechub / Hets

The Heterogeneous Tool Set
http://hets.eu
GNU General Public License v2.0
57 stars 19 forks source link

Can't write "OWL TBox" then "RDF ABox": No inclusion from OWL to RDF found #989

Open sternk opened 10 years ago

sternk commented 10 years ago

Reported by clange and assigned to sternk Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989

On the semantic web (and thus also in OntoIOp) it is common practice to have OWL or RDFS TBoxes, and to write down ABoxes for them in RDF. The semantics of the TBox+ABox union is then best conceived in OWL or RDFS, after translating the RDF ABox to OWL or RDFS. Here is an example for that: https://svn-agbkb.informatik.uni-bremen.de/Hets-lib/trunk/Ontology/Examples/OWLTBoxRDFABox.het – outline:

spec TBoxWithABox =
  TBoxFromSomewhereElse
  then 
  logic RDF : {
    ...
  }

where TBoxFromSomewhereElse is a

spec TBoxFromSomewhereElse =
  logic OWL : {
    ...
  }

On processing the "logic RDF : …" line, Hets says

No inclusion from OWL to RDF found

which is at least somewhat contrary to my intuition. (BTW this problem has the same reason.) However, if there is a different way of telling Hets to

  1. import this TBox in an expressive logic L
  2. then add this ABox in RDF
  3. and treat, whenever necessary, "1 then 2" as "1 then RDFtoL(2)" could you tell me how to do it? Additional question about RDF: When writing down my example, I haven't been able to use Turtle syntax. Are we not yet using a parser that is capable of that? (If not, we should, and I'd be willing to create a separate ticket and an example file.) It seems that just the extremely user-unfriendly N-Triples works at the moment.
sternk commented 10 years ago

Comment by maeder Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:1

Ask Felix for support of the Turtle syntax. Maybe you just need a "cwm" binary (via "make cwm")? The construct "A then B" needs a common super logic. Hets tries to convert the logic of A to the logic of B via an inclusion comorphism (that does not seem to exist for OWL to RDF). Both nodes can be converted manually to a common super logic (maybe CASL), though. Till needs to decide what should be done for this mixture of OWL and RDF.

sternk commented 10 years ago

Comment by clange Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:2

Replying to maeder:

Ask Felix for support of the Turtle syntax. Maybe you just need a "cwm" binary (via "make cwm")?

I had cwm installed (and available in the PATH) before, so that's not the problem. (BTW, I think "make cwm" is not quite user-friendly; see #990). Note once more that here I have RDF inside HetCASL, so I can't use -i format on the Hets command line, but instead I rely on Hets to figure out that my RDF spec is in N-Triples, Turtle, or whatever. (Does Hets currently do this?)

The construct "A then B" needs a common super logic. Hets tries to convert the logic of A to the logic of B via an inclusion comorphism (that does not seem to exist for OWL to RDF).

But here I would like the logic of B to be converted to the logic of A.

Both nodes can be converted manually to a common super logic (maybe CASL), though.

What syntax would convert B to the logic of A? (If possible, I would like to convert from RDF to OWL, not to any other more expressive logic.)

Till needs to decide what should be done for this mixture of OWL and RDF.

OK.

sternk commented 10 years ago

Comment by clange Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:3

Replying to clange:

Replying to [comment:1 maeder]:

Ask Felix for support of the Turtle syntax. Maybe you just need a "cwm" binary (via "make cwm")?

I had cwm installed (and available in the PATH) before, so that's not the problem. (BTW, I think "make cwm" is not quite user-friendly; see #990).

Note once more that here I have RDF inside HetCASL, so I can't use -i format on the Hets command line, but instead I rely on Hets to figure out that my RDF spec is in N-Triples, Turtle, or whatever. (Does Hets currently do this?)

I just reported this as a separate bug.

sternk commented 10 years ago

Comment by clange Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:4

@Till, just a reminder about this. Independently from the ongoing DOL semantics discussion, I consider this quite important for OntoIOp. BTW, I'm also working on a ticket about symbol→term mappings, but it takes some time to get the explanation right.

sternk commented 10 years ago

Comment by clange Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:5

Till, you explained this to me last week, and said that I need to write something like

spec TBox = logic OWL : { ... }
spec ABox =
  TBox hide logic ... %[ what here: "-> RDF", or "RDF2OWL2", or ...? ]%
  then 
  logic RDF : { ... }

Did I get the "hide logic" right? (If so, I will probably have to wait for #957 to be fixed, as RDF2OWL2 is not yet implemented.)

sternk commented 10 years ago

Comment by till Migrated from http://trac.informatik.uni-bremen.de:8080/hets/ticket/989#comment:9

We need a comorphism from OWL to RDF and vice versa.