[georss] place names

[Raj Singh]

(I changed the subject to better reflect this discussion)

Marc, I agree that encoding place names would be a really useful next step,
and Allan's right that the issue can quickly blow up.

But our job shouldn't be to solve the problem of accurate specification of a
place, but adopt a happy medium that works 90% of the time. This is key
because the problem usually blows up on those outlier cases.

So I'll lay out a proposal here.

1. adopt a few tags for place, like:
<address> {street address. As Allan suggests, let developers parse it}
<town>
<region> {this equates to, e.g., state in the US or province in Canada)
<country>
<planet>

2. use full names for town, region and country referenced to some
authoritative source

3. there is no 3!

--Raj


On 6/8/06 4:40 PM, "Allan Doyle" <adoyle at eogeo.org> wrote:

> 
> On Jun 8, 2006, at 16:20, Marc wrote:
> 
>> Hi Everybody
>> 
>> I think we have to be careful not to head too much into the gisRSS
>> direction. Boxes, polygons and other gis entities are certainly
>> useful,
> 
> I'm happy with the set of geometries we have now. I don't see a need
> for more.
> 
>> but the huge majority of rss producers and consumers are playing in a
>> different field. News-feeds, housing-feeds, dating-feeds, job-feeds
>> and
>> so on, all of them would be much more useful if aggregators and
>> rss-reader software could filter, aggregate and group feed items
>> according to 'geo' entities like country code, state, city, etc.
>> With a
>> standard for these encodings geoRSS would make a big step forward, we
>> just have to be careful not to make it too complicated for the
>> mainstream user. RSS is after all a 'Really Simple Syndication'.
> 
> I like the idea of named locations, which is really what an address
> is. I agree that they are more socially useful than geometries.
> 
> Trouble is, addresses are hugely complex. I could envision some kind
> of address blob that can get dropped in. Maybe use the featuretypetag
> and relationshiptag
> 
> Let someone else deal with the address parsing. There are some good
> geocoders out there. Maybe we can convince the geocoder crowd to make
> nice REST interfaces that take an address and produce a GeoRSS geometry.
> 
> There could be some set of Really Useful codes. Country code
> (.de, .ca, etc), Post code (02139, C5G Y0P), Telephone prefix (1,
> 44). The trouble with codes is that they can become out of date
> quickly, too. What will the cc for Montenegro be? How long until IETF
> or ISO catch up and add them? What happens when the code for Serbia-
> Montenegro goes away?
> 
> Maybe we just need some more blobish tags to stuff some interesting
> things into.
> 
> By way of contrast, look at the hoops the microformat people have to
> jump through because XHTML doesn't have anything really suitable.
> 
> Allan
> 
>> 
>> Just my two cents.
>> 
>> Marc
>> _______________________________________________
>> georss mailing list
>> georss at lists.eogeo.org
>> http://lists.eogeo.org/mailman/listinfo/georss
>>