[OSM-talk] Historical Data in OSM database

Laurence Penney lorp at lorp.org
Wed Nov 10 21:25:19 GMT 2010


For the record, I'm 100% against OSM becoming a place for general historical data unless, at the very least, it's been proved that this kind of historical geodata can work well in a parallel database, and shows no sign of interfering with the task of mapping the world as it is. In my first contribution to this thread I listed numerous concerns that including all of history would give rise to, and don't think I came across as supportive.

I'm not sure what you're saying - that 18,000 tag usages is insufficient for someone to try to sort out a mess of tag values?

In my recent message I hoped to sort out the poor usage of a key, about which there's apparently a lot of interest and which in itself is pretty harmless for tagging existing buildings, usefully answering the question "how long has this feature been here?". (Not to denigrate wastebasket mappers, I don't think I'm the only mapper to find such data more interesting than wastebasket locations.) To this end I proposed a general date format that might be potentially useful anywhere that OSM uses human-entered dates.

Could you point to a few occurrences of start_date that refer to "seasonal hours of a tourist attraction only open during the summer" or the like? (I couldn't find any in my review, but I'm using Taginfo which isn't ideal for this stuff.) I thought opening_hours handled this well. In a quick look around I did find some 2010-07-30 start_date tags on the new London cycle hire nodes - seemed reasonable to me, in the historical sense.

- L

On 10 Nov 2010, at 21:40, Richard Weait wrote:

> On Wed, Nov 10, 2010 at 2:51 PM, Laurence Penney <lorp at lorp.org> wrote:
>> It would be good to have consistency in the start_date value. Taginfo reports 18313 usages (2814 distinct), of which these are examples of values other than simple 4-digit years[1]:
> 
> [ ... ]
> 
>> So it's clear there's a demand for: exact dates; general approximations; approximations to the month, season, century; before <date> and after <date>; early <period> and late <period>, maybe also mid <period>; date ranges; multiple values; BC and AD.
> 
> There may be a "clear demand" or interest in start_date, but it is a
> limited one based on your measurement of 18k appearances in the data
> base.
> 
> There are also 18k instances of amenity=waste_basket, [1] 28th in
> amenity and 18k instances of highway=stop [2] 28th in highway.  Both
> waste_basket and stop are clearly defined and are likely to reflect
> only one specific thing.
> 
> By comparison, start_date, may well be used to note the construction
> date or commissioning date of a bridge, but might also define the
> seasonal hours of a tourist attraction only open during the summer.
> Only one of these supports your assertion.  I would argue that
> start_date, for your specific "beginning in an historical sense" use
> is much less prevalent in the data base than you suggest and that
> there is much less "clear demand" for an historical start_date than
> 18,000 appearances might suggest.
> 
> That said, I find the idea of OpenHistoryMap to be a curious idea.  I
> think the idea has potential interest to historians, students,
> developers, genealogists and others.  But I also think it is
> orthogonal to OSM.  If you find the OSM stack helpful in creating
> OpenHistoryMap then do so.  It sounds to me like a Really Big Job
> though.  Not the work of just a weekend.
> 
> But go for it.  Build it based on the OSM stack.  If it can be done in
> a way that keeps OpenHistoryMap contributors happy, and doesn't break
> OSM tools downstream, it might be considered for merging into some
> future OSM.  Even if it does break downstream OSM tools, you'll still
> have a working OpenHistoryMap, and will have had a leg up from
> starting with the working OSM stack.
> 
> [1] http://taginfo.openstreetmap.de/keys/amenity
> [2] http://taginfo.openstreetmap.de/keys/highway
> 
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk




More information about the talk mailing list