From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of Obrst, Leo J.
Sent: Friday, April 18, 2014 8:42 PM
Subject: Re: [ontolog-forum] Reusable Metadata Ontology
If I am reading you right, Andrea, (and please correct me if I’m wrong) the problem behind the problem, i.e., what you are intending to do, is to enable OWL
(ontology, rather than a SKOS lexical vocabulary) to provide you via annotation properties ways to align ontology constructs. Unfortunately in OWL, more complex statements of axioms do have to be put into annotations/documentation.
It would be nice if one could make annotation properties symbolic, i.e., able to be reasoned over. At various times, there were suggestions to make this so.
However, it is difficult to do this (witness the desire to have more complex axioms, as above, which exceed the expressiveness of OWL or description logics more generally), and so it’s fallen aside, as far as I know. There are approaches, in more formally
expressive languages, such as labelled deduction, i.e., so-called hybrid logics, to do that, i.e., correlate additional assertions about provenance, security, belief, etc., i.e., getting into modal logic. I’m not sure whether Common Logic can express this,
but probably IKL can. But OWL can’t.
There have also been attempts to extend the mapping/alignment capabilities of OWL more finely, beyond equivalence statements and import, e.g., epsilon-connections
[1-3] (along the lines of a simple example I gave on the Ontology Summit list earlier this year), but I don’t know that those have been realized. Maybe others can weigh in.
 Kutz, O.; C. Lutz, F. Wolter, M. Zakharyaschev. 2994. E-Connections of Abstract Description Systems, Artificial Intelligence 156 (2004) 1–73.
 Kutz, O. 2004. E-Connections and logics of distance, Ph.D. thesis, University of Liverpool.
 Grau, Bernardo Cuenca; Bijan Parsia; and Evren Sirin. 2006. Combining OWL ontologies using E-Connections. Web Semant. 4, 1 (January 2006), 40-59. DOI=10.1016/j.websem.2005.09.010
On 4/17/14 8:11 PM, Andrea Westerinen wrote:
Leo, I see the confusion. Two things are happening here. As you said in a separate email to me ...
The initial problem is that if you just download any of these ontologies in github (by right clicking
on the file and selecting Save Link As), you get an html file with “github” references everywhere.
If you instead select the file in github just by left clicking on it, and then select the Raw tab, you get the raw file in your
browser window, which you can then save as an apparently true ttl/ofn/owl file. Then you can load it into Protégé.
I am attaching the Turtle version (as a .txt file to get through people's email filters, I hope that it works) to try to avoid the agony of downloading from GitHub without forking.
Let me note also that there are
NO classes, data or object properties in this ontology. There are only
annotation properties that can be used on classes, data and object properties. Since I need this all to be usable in reasoning applications, I started with defining and documenting annotation properties. I try to note this in a comment on the ontology
(but I should probably expand the comment). I am also working on a metadata-properties ontology which defines some of the annotation properties as data and object properties. This will allow (for example) validating dateTime values and referencing objects/individuals
in relations (as opposed to using literal values).
So, for example, I define an exactMatch annotation property that is used to describe how a class or property semantically aligns with another class, property, vocabulary, etc. Again, this has to be done using an annotation property or
I fall into OWL Full. This is obviously totally inadequate to do anything significant, but it is a start as a documentation tool. (There is also a property called closeMatch - as in SKOS - but what really does that mean? How close is "close"?
) Separately, I am working on a more formal approach to mapping but starting with documentation is where I am.
Hope this helps to clarify things,
If you put a Turtle (or any other notation) based RDF document on Github, that's ample for any HTTP user agent to access and consume. The "Raw" page
<https://raw.githubusercontent.com/NinePts/reusable-ontologies/master/metadata/metadata-annotations.ttl> is what exposes your content for direct consumption via an HTTP URL .
I consume the document content directly (we call this sponging) via the publicly available Linked Open Data transformation service that we provide .
As you can see from the UI, there are a lot of raw URIs in the scope of the viewer. You alleviate this problem by adding more rdfs:label, skos:prefLabel etc.. relations i.e., I would have a label and comment (at the very least) for every entity described in
these documents. That's what I do in my glossary of terms doc .
http://linkeddata.uriburner.com/about/html/https/raw.githubusercontent.com/NinePts/reusable-ontologies/master/metadata/metadata-annotations.ttl -- result of basic sponging
http://linkeddata.uriburner.com/describe/?url=""> -- alternative view using a UI with deeper faceted navigation over relations (classes and
instances are in the mix via this UI)
 http://linkeddata.uriburner.com -- Linked Open Data extract, import, and transform (ETL) service
 http://ode.openlinksw.com -- Browser extensions that reduce Linked Open Data ETL and exploitation to a single click or CTRL+Mouse context menu lookup action
 http://bit.ly/19NRwnB -- Glossary of Terms doc
https://github.com/kidehen/GlossaryOfTerms/blob/master/GlossaryOfTerms.ttl -- Github version .
Founder & CEO
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter Profile: https://twitter.com/kidehen
Google+ Profile: https://plus.google.com/+KingsleyIdehen/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen