W3C home > Mailing lists > Public > public-vocabs@w3.org > July 2013

Re: Schema.org extension proposal for job market

From: Jindřich Mynarz <mynarzjindrich@gmail.com>
Date: Tue, 23 Jul 2013 17:34:36 +0100
Message-ID: <CAE=8Bu-k9B9fjqg8AWQUDn+S7rmcdK5oVmc3exVRKgdoVepbFw@mail.gmail.com>
To: "public-vocabs@w3.org" <public-vocabs@w3.org>
Hi all,

I'd like to revive this email thread since the Schema.org extension for job
market [1] presented in it is now more mature. Several modelling choices
were refined and few bugs were fixed. Two new properties were introduced to
the extension. :availableVacancies features the number of persons sought
for in job posting. :employer was created as a super-property of
:hiringOrganization to allow both :Organizations and :Persons to be listed
as subjects offering jobs. The changes are reflected in the proposal's wiki
page [1]. More detailed information is available in the extension's GitHub
repository [2], which now provides comprehensive documentation in English
and Czech [3], [4].

What's also relevant is that we're now aware of the first external use of
the proposed extension. Job-IT [5], a small Czech site for job postings in
IT now contains extended Schema.org markup.

We are now seeking feedback on this updated version, so if you're
interested in the modelling the domain of job market, we'll appreciate your
comments on the proposal.

Finally, I want to ask a question that is relevant to all extension
proposals. When you create new vocabulary terms in your extension proposal,
should you use the http://schema.org/ namespace or should you mint their
URIs in your own namespace (which might be mapped to Schema.org's URI if
the proposal gets accepted)?



[1] http://www.w3.org/wiki/WebSchemas/JobMarket
[2] https://github.com/OPLZZ/data-modelling
[3] https://github.com/OPLZZ/data-modelling/wiki/Reference-English
[4] https://github.com/OPLZZ/data-modelling/wiki/RDFa-Recipe-English
[5] http://www.job-it.cz/

Jindrich Mynarz

On Wed, Feb 20, 2013 at 4:24 PM, Jindřich Mynarz

> Hi,
> the information structures that we included in the proposal are based on
> analysis of corpora of existing job postings and job classifieds (job
> seeker profiles). We looked at a sample of 50 items (postings/classifieds)
> manually and analysed 13 000 items scraped from Craigslist automatically
> (n-grams, word co-occurrence etc.). Based on the samples we distilled the
> most frequently occurring information and got rid of the types of
> information that we thought were either difficult to enter (e.g., ask users
> to maintain the conceptual distinction between "competence" and "skill") or
> not very useful for our primary use case of the data: search and
> match-making.
> Some of the sources we've sampled data from:
> - http://www.adverts.ie
> - http://www.seek.com.au
> - http://chennai.timesjobs.com
> - http://purplecolony.com
> - http://www.usnetads.com
> - http://www.servicejobsite.com
> We plan to use the extended Schema.org vocabulary for a web application
> for job seekers, the first release of which is planned for the end of this
> year. The application will constitute a use case, which will hopefully
> demonstrate the benefits of the additional structures proposed in the
> extension.
> Best,
> Jindrich
> --
> Jindrich Mynarz
> http://keg.vse.cz/resource/person/jindrich-mynarz
> On Wed, Feb 20, 2013 at 4:32 PM, Dan Brickley <danbri@danbri.org> wrote:
>> On 19 February 2013 08:48, Jindřich Mynarz <mynarzjindrich@gmail.com>
>> wrote:
>> > Hi all,
>> >
>> > I've published an early draft of an extension proposal for the
>> Schema.org on
>> > the wiki: http://www.w3.org/wiki/WebSchemas/JobMarket
>> >
>> > The proposal deals with the domain of job market. It adds 2 new classes
>> > (Compensation and WorkExperience) and amends the existing JobPosting
>> class.
>> > More details about the proposal can be found either in the
>> above-mentioned
>> > wiki page or in the GitHub repository containing accompanying diagrams +
>> > RDF/Turtle files.
>> >
>> > Any feedback for making this proposal better is welcome!
>> Thanks for the detailed proposal. Are there any public Web sites that
>> publish this kind of structured data already, which might benefit from
>> adding more markup?
>> Dan
Received on Tuesday, 23 July 2013 16:35:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:29:28 UTC