RDF Resume schema

Hi!

SR> Hi Uldis ... what country are you from?  I am in the USA, located near
SR> Seattle.

I am from Riga, Latvia. Across the globe from you, near Baltic sea.
In #rdfig recently as CaptSolo.

SR> Yes, for now.  That page was kind of a conceptual art of how I want to
SR> progress in the future when I have my bot functional, then  I will fill in
SR> the missing links.

You are working on the bot now? I think I saw you mentioning the bot
somewhere in rdf-interest.

>From the web page it looks to me like MyNetwork is a realization of
FOAF schema or something similar.

I had a similar idea to represent relationships of people in my
address books. Who knows who, who works with who, etc. Agreegate data
from different sources (Palm, cell phone, e-mail software).
Developing further into direction of PIM, export data in
formats that can be imported to each of these devices, keep note of
what addresses appear in (for ex.) cell phone and export only relevant
addresses to it.

Initially I wanted my Master thesis to be about the project above, but
it would have been harder to prove it is usefull to someone else than
me.

>> I am writing a thesis on modelling of personal data, particularly
>> resumes, in RDF (RDFS, DAML+OIL - whichever is best for the purpose).
>> Have you worked in this area and could you give me some pointers to
>> existing resources?

SR> Probably not ... for data identifying the person, it might be nice to use
SR> the foaf namespace:

SR> http://xmlns.com/foaf/0.1/

FOAF will have it's mention/description in my thesis, but as I saw it
pointed out in RDF core discussion of DC with Model Theory that
identifying people by their web pages or by their e-mail address is
not the best way. Similar as it's not the best identifying people by
literal containing their names.

Clearly there is a problem if, let's say, 30% of people submitting CVs
do not have an e-mail address (it might be 1% in other places, and 30%
is just my speculation). FOAF works for information taken from e-mail
directories very well though.

What would be nice in the resume schema, is to provide a rules for translation
between different ontologies so somebody looking at resume could
extract FOAF information from it.

OR - use FOAF (or vCard?) to describe persons information like name,
etc., but something else as ID (URI) for that person.

SR> ... beyond that there is very little agreement on namespaces for persons,
SR> companies, and skills that I know of.

something like person:US:ssn should work if we know ssn.
(although it might be against privacy laws to publish your ssn in a cv)

SR> RDF is a dream for the future ... not enough people have used it yet for it
SR> to have provided the URIs that are necessary for it to be really useful.  If
SR> people don't start looking for URIs for their concepts, and if they cant's

So - people need to use [common] URIs in their RDF data in order for
RDF to get really useful?

Since there are very few ontologies for [in my case] data in resume,
then at start I will have to use significant amount of literals and it will not be
very different from having the same encoded in XML. Then, as new
ontologies get developed, data can be refined to use these ontologies.

SR> find them, make them up, then RDF will not get off the ground.  For the
SR> skills, I assume that there are already existing standards bodies who have
SR> defined skill-names ... converting those to URIs should be straight forward.

What are these standards?

SR> But there is another approach which would still use RDF but not rely so much
SR> on URI.  Use Bnodes instead  ... writing in N3 ...

Blank nodes and use N3?
It will work, but I have to check it as I have not played with
identifying things in terms of description.

Should work for skills, may be used for company names, etc.
The problem persists with identifying the person for whom this CV is.

SR> Look ma, no URIs for XML, RDF, PYTHON and N3.

:) I am acquainted with Java and RDF/XML.
Practical demonstration of my thesis will [have to] be in
Jena, as I do not have time for learining curve to master Python.

Although I have seen interesting things made in Python for SW.
Including Libby's work with conversion of Palm databases to RDF.

SR> At some point somebody might come along and provide URIs:

SR> <uriwhatever:foo/#N3>
SR>     a :Language;
SR>     rdf:label 'N3'
SR>     :identifiedByKeyWord 'N3'
SR>     :seeUrl <http://www.w3.org/2000/10/swap/Primer.html>.

:) Yes, correct, as far as I can tell.

Question: after making this ontology, who will make sure that
.../swap/Primer.html does not move? ... that some company web site (by
which employer is referenced) does not move?

SR> Hmmmm ... I'd think that there would be a direct translation between any
SR> HR-XML data and its corresponding HR-RDF data ... all you would need to do
SR> would be to make a one time XMLT translator.

Would this mean that the only thing I would have to do in my master's thesis,
is to create a "copy" of HR-XML in RDF or to create RDF schema that
supplements HR-XML? That does not sound like very creative thing to
do.

SR> I cc'd Swan Palmer, whom you might already know, on this email, because he
SR> is more adapt at this than I am.

Thanks! I am happy for all the thoughts and suggestions I can get.
Will copy this to www-archive as well.

-- 
Best regards,
 Uldis                            mailto:uldis.bojars@gmx.net

Received on Thursday, 23 May 2002 14:33:49 UTC