Inferred graph vocabulary?

[Gibson, A.P.]

Hi folks,

Does anyone know of, or is perhaps working on, a vocabulary of terms to
describe the relationships between graphs that have been involved in
some reasoning process?

For example, I might take graph <G>, and with ontology <O>, create a
graph of inferred triples <I>, using reasoner R. I then want to add <I>
to my triplestore. I would then want to also specify the provenance of
the inferred graph (from <G> and <O>), and the metadata associated with
its creation (R etc).

This is my particular scenario, I can imagine others. I had thought that
this might fall under the jurisdiction of VoID, but I don't see anything
like that in there (yet). I hope I didn't miss something obvious, I did
Google I promise ;-)

Thanks in advance,
Andrew

--
Dr Andrew Gibson
Universiteit van Amsterdam

<http://www.linkedin.com/in/andrewgibson>

[Richard Cyganiak-2]

Hi Andrew,

VoiD contains terms for expressing subset relationships between graphs  
(we call them datasets), and for listing the vocabularies/ontologies  
that are used in a dataset. The upcoming refined version 2 of voiD  
will allow you to tie void:Datasets to the named graphs in a triple  
store. Furthermore, with the void:Feature mechanism you could define  
features with meaning such as "this graph includes entailed triples  
under RDFS semantics" etc.

This could get you *close* to expressing what you want, but perhaps  
not quite at the level of detail that you want. VoiD is focused a bit  
more on finding/connecting/high-level overview of datasets.

I think the Provenance Vocabulary [1] provides a nice framework for  
the kind of stuff you want to do. You could for example see a reasoner  
as a kind of "data creating service" in the terminology of this  
vocabulary.

There are several publications on "networked graphs" that also seem  
related.

Finally let me ask what's the purpose of publishing/sharing/exchanging  
descriptions of those reasoning processes. If you just use this for  
internal data management, then just define your own internal vocabulary.

Best,
Richard

[1] http://sourceforge.net/apps/mediawiki/trdf/index.php?title=Guide_to_the_Provenance_Vocabulary


On 19 Oct 2009, at 11:04, Andrew Gibson wrote:

> Does anyone know of, or is perhaps working on, a vocabulary of terms  
> to describe the relationships between graphs that have been involved  
> in some reasoning process?
>
> For example, I might take graph <G>, and with ontology <O>, create a  
> graph of inferred triples <I>, using reasoner R. I then want to add  
> <I> to my triplestore. I would then want to also specify the  
> provenance of the inferred graph (from <G> and <O>), and the  
> metadata associated with its creation (R etc).
>
> This is my particular scenario, I can imagine others. I had thought  
> that this might fall under the jurisdiction of VoID, but I don't see  
> anything like that in there (yet). I hope I didn't miss something  
> obvious, I did Google I promise ;-)
>
> Thanks in advance,
> Andrew
>
> --
> Dr Andrew Gibson
> Universiteit van Amsterdam
>
> <http://www.linkedin.com/in/andrewgibson>
>
>

[Giovanni Tummarello]

a) If you dont necessarely need to use RDF to express such
combination, an XML file in the DERI Pipes language will... also be
executable (will fetch, reason, add etc) :-)  http://pipes.deri.ie .
the original need for pipes was indeed flexible combination of
vucabolaries and data

b) Somehow related, you might also be interested in evaluating how
ontologies link each other e.g. starting from a single RDF file, which
is what you see in the ontology tree, calculated live, in the Sindce
inspector. e.g.

http://www.sindice.com/developers/inspector/?url=http%3A%2F%2Frdf4ecommerce.esolda.com%2Fcamcorder%2Fmodel_21107&doReasoning=true#ontologies

(will take a few seconds since its all happening right there)

notice how you get 100 inferred triples out of 70 original ones, use
the triple tab to ese both the original and the inferred (in a
different color)

Hope it helps :-)
Giovanni

On Mon, Oct 19, 2009 at 11:04 AM, Andrew Gibson <a.p.gibson@...> wrote:

> Hi folks,
>
> Does anyone know of, or is perhaps working on, a vocabulary of terms to
> describe the relationships between graphs that have been involved in some
> reasoning process?
>
> For example, I might take graph <G>, and with ontology <O>, create a graph
> of inferred triples <I>, using reasoner R. I then want to add <I> to my
> triplestore. I would then want to also specify the provenance of the
> inferred graph (from <G> and <O>), and the metadata associated with its
> creation (R etc).
>
> This is my particular scenario, I can imagine others. I had thought that
> this might fall under the jurisdiction of VoID, but I don't see anything
> like that in there (yet). I hope I didn't miss something obvious, I did
> Google I promise ;-)
>
> Thanks in advance,
> Andrew
>
> --
> Dr Andrew Gibson
> Universiteit van Amsterdam
>
> <http://www.linkedin.com/in/andrewgibson>
>
>
>

[Leo Sauermann-2]

Hi Andrew,

inferred triples are partly addressed in NRL
it is used in NEPOMUK.

http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/

overview on named graphs which touch inferred graphs a bit
http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/#3._NRL_Named_Graph_Extensions


note that the popular NIE ontologies (used in Nepomuk.kde.org) are built
on top of NRL, so NRL is in production on many linux desktop search
engines in KDE.

best
Leo


It was Andrew Gibson who said at the right time 19.10.2009 12:04 the
following words:

> Hi folks,
>
> Does anyone know of, or is perhaps working on, a vocabulary of terms
> to describe the relationships between graphs that have been involved
> in some reasoning process?
>
> For example, I might take graph <G>, and with ontology <O>, create a
> graph of inferred triples <I>, using reasoner R. I then want to add
> <I> to my triplestore. I would then want to also specify the
> provenance of the inferred graph (from <G> and <O>), and the metadata
> associated with its creation (R etc).
>
> This is my particular scenario, I can imagine others. I had thought
> that this might fall under the jurisdiction of VoID, but I don't see
> anything like that in there (yet). I hope I didn't miss something
> obvious, I did Google I promise ;-)
>
> Thanks in advance,
> Andrew
>

--
_____________________________________________________
Dr. Leo Sauermann       http://www.dfki.de/~sauermann 

Deutsches Forschungszentrum fuer
Kuenstliche Intelligenz DFKI GmbH
Trippstadter Strasse 122
P.O. Box 2080           Fon:   +43 6991 gnowsis
D-67663 Kaiserslautern  Fax:   [phone]
Germany                 Mail:  leo.sauermann@...

Geschaeftsfuehrung:
Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff
Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes
Amtsgericht Kaiserslautern, HRB 2313
_____________________________________________________

[Rinke Hoekstra-4]

Hi Leo, Andrew,

FWIW I think that Andrew's idea to develop a vocabulary for  
indicating, say, 'provenance' of graphs is a very important one. Of  
course, indicating whether some graph is the result of inferencing  
(and by what reasoner) is a very important first step.

Cheers,

Rinke

On 19 okt 2009, at 23:53, Leo Sauermann wrote:

> Hi Andrew,
>
> inferred triples are partly addressed in NRL
> it is used in NEPOMUK.
>
> http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/
>
> overview on named graphs which touch inferred graphs a bit
> http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/#3._NRL_Named_Graph_Extensions
>
>
> note that the popular NIE ontologies (used in Nepomuk.kde.org) are  
> built
> on top of NRL, so NRL is in production on many linux desktop search
> engines in KDE.
>
> best
> Leo
>
>
> It was Andrew Gibson who said at the right time 19.10.2009 12:04 the
> following words:
>> Hi folks,
>>
>> Does anyone know of, or is perhaps working on, a vocabulary of terms
>> to describe the relationships between graphs that have been involved
>> in some reasoning process?
>>
>> For example, I might take graph <G>, and with ontology <O>, create a
>> graph of inferred triples <I>, using reasoner R. I then want to add
>> <I> to my triplestore. I would then want to also specify the
>> provenance of the inferred graph (from <G> and <O>), and the metadata
>> associated with its creation (R etc).
>>
>> This is my particular scenario, I can imagine others. I had thought
>> that this might fall under the jurisdiction of VoID, but I don't see
>> anything like that in there (yet). I hope I didn't miss something
>> obvious, I did Google I promise ;-)
>>
>> Thanks in advance,
>> Andrew
>>
>
>
> --
> _____________________________________________________
> Dr. Leo Sauermann       http://www.dfki.de/~sauermann
>
> Deutsches Forschungszentrum fuer
> Kuenstliche Intelligenz DFKI GmbH
> Trippstadter Strasse 122
> P.O. Box 2080           Fon:   +43 6991 gnowsis
> D-67663 Kaiserslautern  Fax:   [phone]
> Germany                 Mail:  leo.sauermann@...
>
> Geschaeftsfuehrung:
> Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
> Dr. Walter Olthoff
> Vorsitzender des Aufsichtsrats:
> Prof. Dr. h.c. Hans A. Aukes
> Amtsgericht Kaiserslautern, HRB 2313
> _____________________________________________________
>
>

---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law
Faculty of Sciences   |   Faculty of Law
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48
1081 HV Amsterdam     |   1012 CX  Amsterdam
[phone]     |   [phone]
hoekstra@...    |   hoekstra@...

Homepage: http://www.few.vu.nl/~hoekstra

[Leo Sauermann-2]

Hi Andrew,

It was Rinke Hoekstra who said at the right time 20.10.2009 11:20 the
following words:
> Hi Leo, Andrew,
>
> FWIW I think that Andrew's idea to develop a vocabulary for
> indicating, say, 'provenance' of graphs is a very important one. Of
> course, indicating whether some graph is the result of inferencing
> (and by what reasoner) is a very important first step.
As said, this was done in NRL partly, adding one or two properties to
NRL does cover it.
please read NRL and express your additions as a ticket and send it to
oscaf.sourceforge.net,
so that it can get into the implementations.

fyi:
NRL has some wide distribution, an active community with maintiners, a
guarantee to be there the next years with an organisation to back it
(www.oscaf.org), and developers implementing it because they implement
NIE. Its part of nepomuk.kde.org, which is in deployed use on many
desktop pcs in the world. (and will probably be on nokia's maemo platform).

... so there are other people already working on it, you don't have to
start anything new and build your own legal body to own the domain where
you host the ontology for a longer time (if you want, oscaf did that for
you).

this is the decision making process of these ontologies:
https://sourceforge.net/apps/trac/oscaf/wiki/OntologyMaintenance#NRLandNAOMaintenance

an overview of the pending changes to other ontologies:
https://sourceforge.net/apps/trac/oscaf/report/1

The extension you mention is needed for NRL, we are waiting for the
input anyway.

As documented there, the maintainer is Simon Scerri from DERI. I added
him CC.

Of course, doing it properly and documented and tested like this does
take a lot longer, but your input will stay once you have left doing
other things. If you prefer a quick hack you have to maintain yourself,
ignore my mail and do so, I won't bug this topic again in this thread :-)

best
Leo

Andrew, I would suggest to define the extensions you need to NRL and
submit them as a ticket

>
> Cheers,
>
> Rinke
>
> On 19 okt 2009, at 23:53, Leo Sauermann wrote:
>
>> Hi Andrew,
>>
>> inferred triples are partly addressed in NRL
>> it is used in NEPOMUK.
>>
>> http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/
>>
>> overview on named graphs which touch inferred graphs a bit
>> http://www.semanticdesktop.org/ontologies/2007/08/15/nrl/#3._NRL_Named_Graph_Extensions
>>
>>
>>
>> note that the popular NIE ontologies (used in Nepomuk.kde.org) are built
>> on top of NRL, so NRL is in production on many linux desktop search
>> engines in KDE.
>>
>> best
>> Leo
>>
>>
>> It was Andrew Gibson who said at the right time 19.10.2009 12:04 the
>> following words:
>>> Hi folks,
>>>
>>> Does anyone know of, or is perhaps working on, a vocabulary of terms
>>> to describe the relationships between graphs that have been involved
>>> in some reasoning process?
>>>
>>> For example, I might take graph <G>, and with ontology <O>, create a
>>> graph of inferred triples <I>, using reasoner R. I then want to add
>>> <I> to my triplestore. I would then want to also specify the
>>> provenance of the inferred graph (from <G> and <O>), and the metadata
>>> associated with its creation (R etc).
>>>
>>> This is my particular scenario, I can imagine others. I had thought
>>> that this might fall under the jurisdiction of VoID, but I don't see
>>> anything like that in there (yet). I hope I didn't miss something
>>> obvious, I did Google I promise ;-)
>>>
>>> Thanks in advance,
>>> Andrew
>>>
>>
>>
>> --
>> _____________________________________________________
>> Dr. Leo Sauermann       http://www.dfki.de/~sauermann
>>
>> Deutsches Forschungszentrum fuer
>> Kuenstliche Intelligenz DFKI GmbH
>> Trippstadter Strasse 122
>> P.O. Box 2080           Fon:   +43 6991 gnowsis
>> D-67663 Kaiserslautern  Fax:   [phone]
>> Germany                 Mail:  leo.sauermann@...
>>
>> Geschaeftsfuehrung:
>> Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
>> Dr. Walter Olthoff
>> Vorsitzender des Aufsichtsrats:
>> Prof. Dr. h.c. Hans A. Aukes
>> Amtsgericht Kaiserslautern, HRB 2313
>> _____________________________________________________
>>
>>
>
>
> ---
> Dr Rinke Hoekstra
>
> AI Department         |   Leibniz Center for Law
> Faculty of Sciences   |   Faculty of Law
> Vrije Universiteit    |   Universiteit van Amsterdam
> De Boelelaan 1081a    |   Kloveniersburgwal 48
> 1081 HV Amsterdam     |   1012 CX  Amsterdam
> [phone]     |   [phone]
> hoekstra@...    |   hoekstra@...
>
> Homepage: http://www.few.vu.nl/~hoekstra
>
>
>
>

--
_____________________________________________________
Dr. Leo Sauermann       http://www.dfki.de/~sauermann 

Deutsches Forschungszentrum fuer
Kuenstliche Intelligenz DFKI GmbH
Trippstadter Strasse 122
P.O. Box 2080           Fon:   +43 6991 gnowsis
D-67663 Kaiserslautern  Fax:   [phone]
Germany                 Mail:  leo.sauermann@...

Geschaeftsfuehrung:
Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
Dr. Walter Olthoff
Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes
Amtsgericht Kaiserslautern, HRB 2313
_____________________________________________________