Re: data schema / vocabulary / ontology / repositories

Dieter,

Thank you for raising this issue.  I discovered the same problem a couple
years ago, lots of data, no ontologies.
Christopher has a good idea that is not hard to make good progress on. In my
work with linked data, in 08-09, I needed ontologies. So I wrote a simple
automated ontology-extractor. I don't remember the details, but the basic
idea was:
1. create an object property or datatype property for every predicate in
some triple
2. track all the individuals that are used in the subject or object of the
triples, this is a starting point for domains and ranges
3. when individuals are used that are known already (e..g a person in
WIkipedia), classes can be extracted, and this can further information on
domains and ranges.
etc.

I'm sure others have done this kind of thing, and are much more
sophisticated about it.
I did it on a dataset to dataset basis and did not try to use it on multiple
datasets, but it is quite doable.

Michael


On Sun, Mar 13, 2011 at 2:18 PM, Christopher Gutteridge <cjg@ecs.soton.ac.uk
> wrote:

>  That gives me quite an interesting idea.. you could do some studies with
> queries to find what predicates were used to link common classes, e.g. link
> people to documents, to places, to other people...
>
>
> Kingsley Idehen wrote:
>
> On 3/13/11 12:15 PM, Dieter Fensel wrote:
>
> Dear all,
>
> for a number of projects I was searching for vocabularies/Ontologies
> to describe linked data. Could you please recommend me places
> where to look for them? I failed to find a convenient entrance point for
> such
> kind of information. I only found some scattered information here and
> there?
>
> Thanks,
>
> Dieter
>
>
> Dieter,
>
> Do you mean: I would a place where I can search for vocabularies and assess
> their usage across LOD datasets? Goal being reuse of existing terms re. new
> datasets coming into the burgeoning LOD cloud?
>
> If the above is true, the you can do the following:
>
> 1. Goto http://lod.openlinksw.com  -- the live 15 Billion+ triples LOD
> Cloud Cache we maintain
> 2. Enter a text pattern (with Class, Property, or Vocabulary discovery in
> mind)
> 3. On receipt of the initial results page, use the Links in the Navigation
> section to filter by Type or other Attributes (so you are looking for
> Entities of type: Ontology or Class or Property
> 4. Once you find one of the Entity Types above, click on the "describe"
> link
> 5. At this point navigation should be obvious i.e. you can use isDefinedby
> to find the Ontology associated with Classes and Properties or use the
> inverse relations to find the Class and Properties defined by an Ontology.
>
> Hope this helps.
>
> --
>
> Regards,
>
> Kingsley Idehen	
> President & CEO
> OpenLink Software
> Web: http://www.openlinksw.com
> Weblog: http://www.openlinksw.com/blog/~kidehen
> Twitter/Identi.ca: kidehen
>
>
>
>
>
>
>
> --
> Christopher Gutteridge -- http://id.ecs.soton.ac.uk/person/1248
>
> You should read the ECS Web Team blog: http://blogs.ecs.soton.ac.uk/webteam/
>
>


-- 
Michael Uschold, PhD
   Senior Ontology Consultant, Semantic Arts
   LinkedIn: http://tr.im/limfu
   Skype, Twitter: UscholdM

Received on Tuesday, 15 March 2011 13:21:03 UTC