[Data-modeling] historical locations

[Jeff Prucher]

These are really good questions, and it's definitely a difficult modeling
problem.  

One approach might be the one taken on the "company" type
(http://www.freebase.com/view/schema/business/company), which might get us
partway there. Briefly, it has three pairs of properties: spun off from/spun
off; acquired by/companies acquired; involved in merger/product of merger.
These pretty much sum up the way companies can become other companies.
Locations would necessarily have a different set of properties, but this
general approach might work well.  I'm going to defer actually sorting out
what those properties could or should be for the time being, just due to the
other things on my plate.  If someone else wants to tackle this sooner,
they'd be more than welcome!

Re postal codes, we can definitely add "dated location" as an included type.

Speaking of which, I'll promote the dated location type (as it stands) later
this week unless I hear objections.

Thanks,
Jeff Prucher






> -----Original Message-----
> From: data-modeling-bounces at freebase.com 
> [mailto:data-modeling-bounces at freebase.com] On Behalf Of Tom Morris
> Sent: Wednesday, November 28, 2007 5:57 PM
> To: Freebase data modeling mailing list
> Subject: Re: [Data-modeling] historical locations
> 
> How are changing bounds, names, and administrative 
> hierarchies dealt with?  The town of Dunstable, MA makes an 
> interesting case to see if you can encode.  It's original 
> bounds spanned the border of present day MA and NH.  At some 
> point the border was redrawn by the King of England, creating 
> Dunstable, NH and Dunstable, MA.  Over time, Dunstable, NH 
> was renamed several times (but kept the same bounds), 
> eventually becoming Nashua, NH.  Dunstable, MA kept it's 
> name, but got smaller and smaller as additional towns were 
> carved out of it.
> 
> Note that this isn't just of academic interest.  If you were 
> tracing the provenance of a deed for a piece of land that 
> changed jurisdictions, you'd need to know when to look where.
> 
> The historical translations need to take place in both 
> directions.  If you wanted to visit Ye Olde Dunstable (MA) 
> Burying Ground, you'd find it in present-day Nashua, NH.
> 
> Tom
> 
> p.s.  The earlier discussion mentioned not having to deal 
> with expiring postal codes.  The zip code that I grew up in 
> is no more.
> I'm not sure exactly when or why, but at some point it was 
> replaced and it's no longer a valid zip, so if you end up 
> with a historical address containing the zip 02181, you may 
> want to know that it was valid at some point even though it 
> isn't now and also what it translates to in the present day 
> (or some other point in history).
> 
> On Nov 28, 2007 6:54 PM, Jeff Prucher <jeff at metaweb.com> wrote:
> > I've finally gotten around to taking a stab at this.  (For 
> reference, 
> > the complete discussion thread is archived here:
> > 
> http://lists.freebase.com/pipermail/data-modeling/2007-October/thread.
> > html.) I've created a type called "dated location", which has two 
> > properties: "date formed" and "date dissolved".  This is 
> intended to 
> > be used as an included type on /location/country, 
> > /location/administrative_division,
> > /location/citytown, /protected_sites/protected_site, and 
> > /government/political_district, but I haven't added it 
> there yet (the 
> > type is still in my private domain), although it could obviously be 
> > added to any topic.
> >
> > Please take a look at the type and let me know what you think. I've 
> > put in some data for various incarnations of Hawaii as a test set.
> > http://www.freebase.com/view/user/jeff/default_domain/dated_location
> >
> > Thanks,
> > Jeff P
> >
> > > -----Original Message-----
> > > From: data-modeling-bounces at freebase.com
> > > [mailto:data-modeling-bounces at freebase.com] On Behalf Of Jeff 
> > > Prucher
> > > Sent: Monday, October 22, 2007 10:33 AM
> > > To: 'Freebase data modeling mailing list'
> > > Subject: Re: [Data-modeling] historical locations
> > >
> > > The big question here is whether we need a new 
> "historical location" 
> > > type (parallel to the "deceased person" and "defunct 
> company" types, 
> > > or whether slapping a "date dissolved" (or equivalent 
> property) on 
> > > the current relevant types (country, administrative division, and 
> > > city/town) would do the trick.
> > > A "date formed" property should really be on these base 
> types, since 
> > > current (i.e., non-historical) locations also have this value.
> > >
> > > Jeff P
> > >
> > > > -----Original Message-----
> > > > From: data-modeling-bounces at freebase.com
> > > > [mailto:data-modeling-bounces at freebase.com] On Behalf 
> Of Ryan Shaw
> > > > Sent: Sunday, October 21, 2007 5:22 PM
> > > > To: Freebase data modeling mailing list
> > > > Subject: Re: [Data-modeling] historical locations
> > > >
> > > > On Oct 21, 2007, at 5:05 PM, Kirrily Robert wrote:
> > > >
> > > > > I feel like there should be a type, "Historical 
> location", which 
> > > > > includes type "location" and then has start/end dates.
> > > Has anyone
> > > > > worked on this already?
> > > >
> > > > Yes, this is needed. There are a lot of historical 
> gazetteers that 
> > > > record the historical dates when particular place names
> > > were in use--
> > > > if we ever want to represent this kind of info in Freebase
> > > we need a
> > > > historical location type. In fact I am hoping to import
> > > some lists of
> > > > historical Irish place names, from the Onomasticon
> > > Goedelicum, at some
> > > > point...
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
>