[rdfweb-dev] Re: [foafcorp-dev] The new hierarchies site is in beta

James Moore jam at wirerimmed.com
Sun Sep 21 20:12:51 UTC 2003


On Saturday, September 20, 2003, at 04:52  PM, Dan Brickley wrote:

> * James Moore <jam at wirerimmed.com> [2003-09-19 19:41-0700]
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> I've just posted the new version of the Hierarchies Project website
>> [1]. Right now it will produce RDF [2] and SVG [3] versions of company
>> listings. The schema I'm using is what was batted around briefly 
>> *many*
>> months ago so I'd appreciate anyone interested in FOAFCorp to glance 
>> at
>> it.
>
> Cool! I've just started a little experiment in using FOAF to support
> journalism, so was looking for an fc: way of expressing 'nasdaq
> stockticker' just now. Do you have that? For now I just made up a bs
> property in the current fc namespace, calling it 'nasdaq' while not
> really yet understanding how many different stockticker ID value spaces
> there are. If just one, then 'stockID' might have been a better name.
>

i've added ticker symbol, common name, and notes to the DB schema. The 
RDF schema hasn't been updated to reflect this yet. Obviously I'd like 
to reuse definitions wherever possible. I found a reference to some 
foafcorp like work [1] that not only lists the symbol but also the 
market. I'd be happy to just have ticker symbol and assume nasdaq to 
start with since most of our data is for US companies.


> Looking at your 2.rdf, a question: what does the 'Relationship' 
> property
> mean? Just is just hold between companies? Does it mean something like
> 'corporate alliance', or 'owns', 'is owned by', 'subcontracts to' etc?
>

Relationship is just a way of categorizing a parnet-child link between 
2 entities. Right now an entity is a corp or NGO. For example it could 
define a company->division link or a lobby group->member link. There's 
definitely room for more categories like 'subcontracts to'.

> foaf:owns was suggested on rdfweb-dev recently btw., and discussed a
> little in #foaf IRC. It would have a range of foaf:Agent (which would
> include fc:Company), and a domain of 'Resource' (ie. everything is
> considered ownable). Not sure how that fits w/ foafcorp extensions, nor
> where its best to draw the line between the two.
>

I'm making a distinction between a corp owning a brand and owning 
another company. So Assets (like brands)  are coupled directly to the 
Entity. A corp that's owned by another corp is linked via an 
intermediate property that can itself have some properties, percentage 
of ownership for example.

The approach I took to relationships looks like:

<rdf:Description rdf:about="#E2">
<rdf:type rdf:resource="http://hierarchies.org/rdf/schema#Entity"/>
<foaf:homePage rdf:resource="http://www.philipmorris.com"/>
<foaf:name>Altria Group Inc.</foaf:name>
<hp:category>Corporation</hp:category>
<contact:Address> New York, NY US</contact:Address>
<foaf:phone rdf:resource="tel:917-663-5000"/>
<hp:Relationship rdf:resource="#R43"/>
</rdf:Description>


<rdf:Description rdf:about="#R43">
<hp:category>Subsidiary</hp:category>
<hp:child rdf:resource="#E60"/>
<hp:percentage>100</hp:percentage>
</rdf:Description>

<rdf:Description rdf:about="#E60">
<rdfs:seeAlso rdf:resource="http://beta.hierarchies.org/rdf/60.rdf"/>
</rdf:Description>

If the entity was a board then the relationship could 'board member' 
and the child could point to the foaf for the board member. I defined 
the range of child to be an entity but i think now that was a bad idea.

> (time passes, I get distracted by http://www.gregpalast.com/ and adding
> to my FOAF/journo demo. Oh yeah, that's what I was going to say)
>
> So...
>
> 	http://rdfweb.org/2003/09/mirrors/jetblue/jetblue.rdf

> ...is an example of some RDF/XML using FOAF and other vocab (foafcorp, 
> DC,
> wot, rdfs) to describe some raw materials of journalistic/activistic
> interest. In this case relating to the Company 'Jet Blue' and their
> recent naughtyness w.r.t. donating data to creepy TIA-style research.
>

One thing that isn't expressed in our current schema is what we refer 
to as Effects and Source. For example Dyncorp has been in some trouble 
in the past [2] and we *really* want people to submit that kind of 
info. If the schema included this data then your jet blue example could 
become somewhat smaller by just linking to the main jet blue rdf 
document on hierarchies right? I realize the metadata for the source is 
rather slim and would need some beefing-up before it could really work 
the same way.

>
> OK, rambling now. I might try write this up for the weblog.
>
> Re Foafcorp and hierarchies schema and Zool's schema, how best to
> proceed? I'm totally up for converging these, and (if not too
> presumptious) happy to offer the FC namespace at
> http://xmlns.com/foaf/corp# for the results of any such collaboration,
> merging. Zool has CVS access now, and you'd be welcome to it too. We
> could invent some change-control policies, like those used for
> http://esw.w3.org/topic/RdfCalendar / http://www.w3.org/2002/12/cal/
> perhaps).
>

Using the foafcorp namespace for the hierarchies data is fine by me. 
Right now there's almost no overlap between our two schemas though. 
What would be the drawbacks of adopting the current hierarchies schema 
and moving forward with changes from there? The only reason I suggest 
this is that hierarchies has usable data right now even if the schema 
gets changed later.


> From the rdf-calendar page:
> [[
> At the Bristol workshop, we agreed, roughly...
>
>         * we announce all changes to the schema www-rdf-calendar
>         * if anyone screams, within a week or so, we'll back out the
> 	* changes (for further discussion)
>
> 	If the last modified dates below are more than a few months old,
> 	active developments have likely ceased. If things ever actually
> 	stabilize, we'll change this status message.
> ]]
>
> Maybe we could do same w/ foafcorp vocab?

Sounds good to me.

-James



[1] http://web.resource.org/rss/1.0/modules/company/
[2] http://beta.hierarchies.org/entity/7/effects




More information about the foaf-dev mailing list