W3C home > Mailing lists > Public > semantic-web@w3.org > March 2008

Re: meta-information about assertions

From: Richard Cyganiak <richard@cyganiak.de>
Date: Tue, 4 Mar 2008 15:55:24 +0000
Cc: semantic-web@w3.org
Message-Id: <AE4CF756-1B71-44F8-8D69-40281A5924D0@cyganiak.de>
To: Cristiano Longo <cristiano_longo@yahoo.it>


On 4 Mar 2008, at 15:21, Cristiano Longo wrote:
> Let me summarize. There is too main trends.
> The first(Phil Archer, Story Henry) suggests to keep
> track of a "graph" and record graph provenance. If I
> understood good, it means to use QUADS or to put into
> my knowledge base only graph uri and graph provenance.

Not quite. The idea is that RDF statements never exist just floating  
around freely; they are always inside some container, a graph or a  
document. Now if that graph or document also has a URI as its name,  
then we can use that URI to make statements about the container, e.g.  
who created it, at what time, for what reason etc. And implicitly,  
this metadata applies not just to the container but also to all  
statements contained in it.

To store RDF using this model, you need a quad store, not just a plain  
triple store. All modern RDF stores are quad stores.

SPARQL is based on this model.

> The second trend is to store assertions into rdf using
> something like realization

You mean reification?

> , creating for example an
> individual assertion with an asserter and some
> property
> filled with the assertion.

The main difference is that with reification, we give a URI as a name  
to each individual RDF statement. Then we can make statements about  
each individual RDF statement. This is more flexible than the Named  
Graphs model, but it makes management of statements much more complex.

On the public Semantic Web, you will find lots of RDF documents that  
contain metadata or statements about other RDF documents (Named Graphs  
style), but there's very little public RDF data that uses reification.  
Even though reification has been around much longer.

> I noticed that Evaluation and Report Language,
> Semantic Web Publishing Vocabulary(WIQA) and Ratings
> Ontology follows this second trend(probably also Proof
> Markup Language), providing properties to specify the
> asserter of something(e.g. in EARL an agent assert
> that a document passed or not a checkpoint).

Uhm, WIQA is in the first camp, it's based on the Named Graphs model,  
like SPARQL. Both approaches need properties to specify e.g. the  
asserter of something with both approaches, that's not a difference.  
The difference is that the first approach attaches metadata to entire  
containers of RDF statements, while the second approach attaches  
metadata to individual RDF statements.

Best,
Richard


>
>
> So i think that probably(if not exists), for
> interoperability purposes, we need to formalize these
> situations into a main specification(or extending a
> suitable one), preferably grounding it with
> description logics.
>
> Thank you and sorry for my english,
> Cristiano Longo
>
> --- Chris Bizer <chris@bizer.de> ha scritto:
>
>> Hi Cristiano,
>>
>> as Steffen already said, using the Named Graph data
>> model together with
>> SPARQL is a  practical and well tested way for doing
>> this.
>>
>> Reification is considered dead by most people
>> working on Semantic Web-based
>> data integration. I think the only people still
>> thinking about using
>> reification are the new OWL working group and I hope
>> that they will also
>> realize at some point that they are running into
>> problems with this.
>>
>> If you need a vocabulary for representing
>> meta-information about graphs, one
>> option is to use the Semantic Web Publishing
>> vocabulary. A framework that
>> might be interesting for you with regards to trust
>> is the WIQA Web
>> Information Quality Assessment framework, which
>> employs the Named Graphs
>> data model and allows you to formulate various
>> information filtering
>> policies using a policy language that is based on
>> SPARQL.
>>
>> See:
>>
>> http://www4.wiwiss.fu-berlin.de/bizer/WIQA/index.htm
>>
> http://www4.wiwiss.fu-berlin.de/bizer/WIQA/browser/index.htm
>>
>> Cheers
>>
>> Chris
>>
>>
>> --
>> Chris Bizer
>> Freie Universitšt Berlin
>> +49 30 838 54057
>> chris@bizer.de
>> www.bizer.de
>> ----- Original Message -----
>> From: "Steffen Staab" <staab@uni-koblenz.de>
>> To: "Cristiano Longo" <cristiano_longo@yahoo.it>
>> Cc: <semantic-web@w3.org>
>> Sent: Tuesday, March 04, 2008 9:13 AM
>> Subject: Re: meta-information about assertions
>>
>>
>>> Hi,
>>>
>>> here is a WWW08 paper about this:
>>>
>>
> http://www.uni-koblenz.de/~staab/Research/Publications/2008/WWW2008-MetaKnowledge.pdf
>>> and here is its implementation:
>>>
>>
> http://www.uni-koblenz.de/FB4/Institutes/IFI/AGStaab/Research/MetaKnowledge
>>>
>>> Cheers,
>>> Steffen
>>>
>>> Cristiano Longo schrieb:
>>>> Hi all,
>>>> i'm trying to merge rdf(more specifically OWL)
>> graphs
>>>> from different sources using collaborative
>> filtering
>>>> and trust related technologies. But my question
>> is:
>>>> what is the proper way to encode a "meta
>> assertion"
>>>> like "A says X about B", in order to deal with
>>>> contraddictory assertions?
>>>>
>>>> Reification? Using SKOS? Something else?
>>>>
>>>> Thank you in advance.
>>>>
>>>>
>>>>      ___________________________________
>>>> L'email della prossima generazione? Puoi averla
>> con la nuova Yahoo! Mail:
>>>> http://it.docs.yahoo.com/nowyoucan.html
>>>>
>>>>
>>>>
>>>
>>
>>
>
>
>
>      ___________________________________
> L'email della prossima generazione? Puoi averla con la nuova Yahoo!  
> Mail: http://it.docs.yahoo.com/nowyoucan.html
>
Received on Tuesday, 4 March 2008 15:55:45 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:05 UTC