I would like to know if there exists a formal methodology to map an ontology to a set of rules (RIF or SPARQL-based rules such as SPIN Rules) that could be used to validate instances of classes of a given ontology. I am not interested by validation tools such as Pellet ICV or SPIN engine, but by their underlying specification validation rules. Is it possible to have a complete validation rule set to cover all constructs of OWL ? The following link should clarify what I am looking for: http://clarkparsia.com/pellet/icv/

asked 28 Feb '13, 16:10

fellahst's gravatar image

fellahst
3.3k29
accept rate: 11%

edited 06 Mar '13, 12:05

Signified's gravatar image

Signified ♦
24.0k1623

2

What do you mean by "validate instance of classes"? If you want to translate the complete OWL semantics into rules, then good luck. You'd need complicated rules, unless you can determine that the ontology is in a fragment like OWL 2 RL or OWL Horst.

(01 Mar '13, 02:57) Antoine Zimm... ♦ Antoine%20Zimmermann's gravatar image

The following link should clarify what I am looking for: http://clarkparsia.com/pellet/icv/

(01 Mar '13, 10:17) fellahst fellahst's gravatar image

Assuming the ontology is encoded in OWL2 then there are SPIN rules that check that the ontology is consistent according to OWL2-RL. OWL (e.g. full) has certain profiles that are not encodable in decidable rules i.e. are not guaranteed to terminate. For example your ontology data might be consitent under the RL profile but not the DL profile or the other way round. As all of OWL is not encodable in rules, then no it not possible to have a complete (assuming OWL-full) validation rule set for OWL. It would be possible for OWL2-RL but it also depends on what exactly is included in "complete validation rule set".

permanent link

answered 01 Mar '13, 11:07

Jerven's gravatar image

Jerven ♦
4.6k610
accept rate: 34%

I understand that not all of OWL can be encoded as rules. Your answer claims that there are SPIN rules that check the OWL_RL ontology. What are these rules ? Is it based on some standards ? That is what I am looking for.

(01 Mar '13, 11:20) fellahst fellahst's gravatar image
4
(04 Mar '13, 18:44) Holger Knubl... Holger%20Knublauch's gravatar image

@Holger Knublauch thanks, I was hoping you would have time to add this kind of information.

(05 Mar '13, 13:53) Jerven ♦ Jerven's gravatar image

Are you looking for the formal semantics behind Pellet ICV? If so: http://stardog.com/docs/sdp/icv-specification.html

permanent link

answered 05 Mar '13, 15:18

mhgrove's gravatar image

mhgrove
3.3k17
accept rate: 29%

Yes, indeed. Thank you for the link. Are you aware of any translation in RIF or SPARQL of these rules ?

(06 Mar '13, 09:49) fellahst fellahst's gravatar image
1

Yes, the Implementing IC Semantics discussion describes both type of translation. Primarily we use the SPARQL translation.

(06 Mar '13, 09:53) mhgrove mhgrove's gravatar image

Your question sounds a little confused so just to clarify the key issue: Pellet ICV considers a closed world assumption (CWA). OWL considers an open world assumption (OWA). This is a thorny issue with a few related questions on this site (for example, here's one and I have a related answer here), but suffice it to say that Pellet ICV is doing something that is incompatible with standard OWL and that will not be supported by standard OWL reasoners/profiles.

===============

Maybe a brief example to clarify:

 :hasChild rdfs:domain :Parent .
 :Fred :hasChild :Mary .

Under OWA (and RDFS or OWL), :Fred a :Parent is considered missing, and so can be added.

Under CWA (and probably Pellet ICV), :Fred a :Parent is considered untrue, and so its omission is an error.

Domain and range act as constraints under CWA but not under OWA.

===============

OWA can also have constraints. For example:

:Person owl:disjointWith :Cabbage .
:Fred a :Person , :Cabbage .

Under CWA or OWA, this is a clear contradiction.

===============

On a side note, you also might in interested in an draft W3C specification for using RIF to support the OWL 2 RL/RDF rules. Again, this only supports an OWA and the second type of contraints above. However, on top of that, you can use RIF to support CWA-style constraints using additional rules.

permanent link

answered 06 Mar '13, 12:00

Signified's gravatar image

Signified ♦
24.0k1623
accept rate: 37%

edited 06 Mar '13, 12:03

1

Oh! :hasChild rdfs:domain :Parent . :Fred :hasChild :Mary . implies that :Fred a :Parent, no matter CWA or OWA. However, under CWA, :Fred a :Father is considered untrue.

(07 Mar '13, 01:56) Antoine Zimm... ♦ Antoine%20Zimmermann's gravatar image
Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Question tags:

×668
×69
×29
×2
×1

question asked: 28 Feb '13, 16:10

question was seen: 1,336 times

last updated: 07 Mar '13, 01:56