> I've recently found UMBEL. From what I read, it allows the search for
> the contextualization of concepts of external ontologies.
>
Well, umbel can be used for many purposes; but it is right that one of
the main purposes is to link external ontologies classes together by
using a common framework (umbel). Then this enable inference on classes
of external ontologies that was not possible before their linkage to
umbel because of the missing subsumption links.
> I have seen that there are various web services that allow search on
> umbel. My question is, and excuse my ignorance, associations between
> external concepts and Umbel concepts hasn't no way to infer by not
> being part of the web version of umbels. Is this correct?
>
I am not sure I understand your questions, but: linkage between external
ontologies classes and umbel classes (subject concepts) is available in
this file [1]. So basically, you can load the umbel subject concept
structure in your triple store, load that file, and then you will be
able to do inference on these classes.
> I read a post in which the management of Umbel occurs locally
> (downloading and managing them by Jena Framework) is this the right
> way to work with UMBEL? What is the best way to work with the shared
> web version? (I consider the web accessibility is a really important
> element of UMBEL).
>
Well, you have two choices:
(1) you load UMBEL & anything else you could need in your local
datastore (Jena if it is what you want to use)
(2) you use the UMBEL Web Services for a "plug-and-play" version of UMBEL.
> Abusing a little :), if the management should be local, is the
> framework of Jena sufficient or it is preferable to use other
> components as Virtuoso server, ...?
>
Well, I personally didn't tested UMBEL on Jena but I think some people
did. Otherwise the Web Services are using Virtuoso, and the other
projects I am working on also uses Virtuoso. So, what I can say is:
Virtuoso do the job; we would have to test with other data stores to see
if they are well suited too :)
> Thank you very much in advance.
>
Thanks to you!
[1] http://umbel.org/ontology/external_ontologies_descriptions.zip
Take care,
Fred
> How are you crafting the RDF to insert into Virtuoso? Are you using
> Protege or some other platform?
>
We have scripts that take new OpenCyc versions as input, index it into
the triple store, then that create the umbel SCs, ACs and NEs from that
dataset. We also have list of candidates that we use to know what will
be included and what will be ignored when we will produce these
different datasets. Additionally, there are all the sanitization
procedures that ensure that new structure we are creating is consistent,
that there are no gaps in the OpenCyc subset, that the linkage with
external classes is still consistent, and that we are not missing new
concepts.
Thanks!
Fred