Re: [OSM-talk-be] Addresses in Belgium

[Kurt Roeckx]

On Tue, Jan 08, 2013 at 12:02:07AM +0100, Jo wrote:
> 2013/1/7 Kurt Roeckx 
> 
> > On Mon, Jan 07, 2013 at 06:06:02PM +, Jan-willem De Bleser wrote:
> > > Of course, that leads me to the question of "Why do
> > > we add addr:city at all, assuming that every house is fully within a
> > > city's boundaries?"
> >
> > The only reason I can see that being useful is that border between
> > 2 cities goes through the building.
> >
> 
> The reason why I want to add addr:city and addr:postcode via an
> associatedStreet relation is that not all data consumers have geographic
> databases at their disposal

That they can't extract the data doesn't mean we need to duplicate
it.  In general it's a bad idea to have multiple sources of the
same information, since they will conflict at some point.

> and Openstreetmap doesn't have an API (yet)
> that allows you to query what city an object is in.

Nominatim gives you an xml file that includes the city.


Kurt


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

2013/1/7 Kurt Roeckx 

> On Mon, Jan 07, 2013 at 06:06:02PM +, Jan-willem De Bleser wrote:
> > Of course, that leads me to the question of "Why do
> > we add addr:city at all, assuming that every house is fully within a
> > city's boundaries?"
>
> The only reason I can see that being useful is that border between
> 2 cities goes through the building.
>

The reason why I want to add addr:city and addr:postcode via an
associatedStreet relation is that not all data consumers have geographic
databases at their disposal and Openstreetmap doesn't have an API (yet)
that allows you to query what city an object is in.
It is trivial though to ask what associatedStreet relation does an object
belong to and get the complete address that way.

Jo
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Kurt Roeckx]

On Mon, Jan 07, 2013 at 06:06:02PM +, Jan-willem De Bleser wrote:
> Of course, that leads me to the question of "Why do
> we add addr:city at all, assuming that every house is fully within a
> city's boundaries?"

The only reason I can see that being useful is that border between
2 cities goes through the building.


Kurt


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 5:20 PM, Sander Deryckere  wrote:
> If it's administrative, it's not necessarily the closest. I have given an
> example of it, but there are multiple examples, like for this house:
> http://www.openstreetmap.org/browse/way/174076563, it's in gemeente Staden,
> but all streets around it are in gemeente Roeselare. So you can't add those
> streets to the associatedstreet relation if it's administrative. Unless all
> driveways should also be part of the relation.

So, this house has the official address "Groenestraat 42, Staden", but
the driveway opens on "Groenestraat, Roeselare"? And the problem is
that if you add it to the associatedStreet, the house will look like
it's in Roeselare, right?

That is tricky, but the problem is not the relation linking the street
and the house, but that the relation is tagged as being in Roeselare.
Or have I misunderstood your explanation?

>> Or are you all really only interested in associatedStreet as a
>> gathering point for the common information such as postcode and
>> country?
>
> I thought so, as it's a general recommendation to not use relations where
> spatial queries can be 100% accurate. Administrative stuff can't be 100%
> accurate queried, closest street can.

'Closest street' is 100% accurate, but the street a house belongs to
is not necessarily the closest one, especially if you were to consider
a street that crosses a town boundary to be two separate streets. To
take a similar problem, if you have a house that is divided in half by
a postcode boundary, how do you determine in which postcode the house
belongs? This is the kind of problem I had in mind when I said, in my
first mail, that such a search was "not necessarily accurate".

I do see your point that you shouldn't tag what you can find in a 100%
accurate search. Of course, that leads me to the question of "Why do
we add addr:city at all, assuming that every house is fully within a
city's boundaries?"

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Sander Deryckere]

On 7 Jan 2013 17:51, "Jan-willem De Bleser"  wrote:
>
> On Mon, Jan 7, 2013 at 5:07 PM, Jo  wrote:
> > So all the ways forming a street with addresses in the same city,
having the
> > same postcode, together with all the houses go into the same
> > associatedStreet relation.
>
> Ah, ok, one relation containing all the ways and all the houses - I
> agree, but please don't forget my mail from 4:36. You will indeed
> still need to perform a spatial search, but this of a subset of all
> data in the immediate neighborhood, and only in the subset of way
> objects guaranteed to represent the correct street.

If it's administrative, it's not necessarily the closest. I have given an
example of it, but there are multiple examples, like for this house:
http://www.openstreetmap.org/browse/way/174076563, it's in gemeente Staden,
but all streets around it are in gemeente Roeselare. So you can't add those
streets to the associatedstreet relation if it's administrative. Unless all
driveways should also be part of the relation.
>
> Or are you all really only interested in associatedStreet as a
> gathering point for the common information such as postcode and
> country?

I thought so, as it's a general recommendation to not use relations where
spatial queries can be 100% accurate. Administrative stuff can't be 100%
accurate queried, closest street can.
>
> - Jw
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-be
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 5:07 PM, Jo  wrote:
> So all the ways forming a street with addresses in the same city, having the
> same postcode, together with all the houses go into the same
> associatedStreet relation.

Ah, ok, one relation containing all the ways and all the houses - I
agree, but please don't forget my mail from 4:36. You will indeed
still need to perform a spatial search, but this of a subset of all
data in the immediate neighborhood, and only in the subset of way
objects guaranteed to represent the correct street.

Or are you all really only interested in associatedStreet as a
gathering point for the common information such as postcode and
country?

- Jw

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Kurt Roeckx]

On Mon, Jan 07, 2013 at 08:17:11AM +0100, Jo wrote:
> The associatedStreet relation has the streetname in 'name', not in
> addr:street. I also found some relations where this was done incorrectly.
> 
> It is possible to fix all of them in one go. Advise me if you want me to do
> so.

As far as I understand it, there is no reason to give the relation
a name other than being useful to find it.  The street name itself
should come from the member that is marked as street.  It something
displays the name of the relation as addr:street information, I think
that's just wrong and should get fixed.


Kurt


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

2013/1/7 Jan-willem De Bleser 

> On Mon, Jan 7, 2013 at 4:49 PM, Sander Deryckere 
> wrote:
> > Who says that the closest street is in the associatedstreet relation.
> That
> > relation has nothing to do with the closest street, only with the
> > administrative division of houses into streets.
> >
> > Look at this relation:
> http://www.openstreetmap.org/browse/relation/1869108
> >
>
> Hang on, by "closest street" you mean "closest way of a street mapped
> as multiple ways"? It's my understanding that, when this is the case,
> a house is associated with the way on which it lies. That relation
> 1869108 is an example of incorrect mapping, as far as I can see.
>
> Addresses are associated with a particular stretch of street, aren't
> they? I've always taken associatedStreet as a relation trying to
> represent this mapping. Or would you maintain that this is true, but
> that the stretch of street belonging to an address bears no relation
> to where the plot of land belonging to that address is?
>

At first the definition of associatedStreet was like you say, but this has
been changed. It's too hard to keep it correct (when splitting ways for
example).

So all the ways forming a street with addresses in the same city, having
the same postcode, together with all the houses go into the same
associatedStreet relation.

BTW, there is a great mapcss for JOSM called ColouredAddresses which give a
great overview of what belongs together.

Jo
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 4:49 PM, Sander Deryckere  wrote:
> Who says that the closest street is in the associatedstreet relation. That
> relation has nothing to do with the closest street, only with the
> administrative division of houses into streets.
>
> Look at this relation: http://www.openstreetmap.org/browse/relation/1869108
>

Hang on, by "closest street" you mean "closest way of a street mapped
as multiple ways"? It's my understanding that, when this is the case,
a house is associated with the way on which it lies. That relation
1869108 is an example of incorrect mapping, as far as I can see.

Addresses are associated with a particular stretch of street, aren't
they? I've always taken associatedStreet as a relation trying to
represent this mapping. Or would you maintain that this is true, but
that the stretch of street belonging to an address bears no relation
to where the plot of land belonging to that address is?

- Jw

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Sander Deryckere]

Who says that the closest street is in the associatedstreet relation. That
relation has nothing to do with the closest street, only with the
administrative division of houses into streets.

Look at this relation: http://www.openstreetmap.org/browse/relation/1869108

So you end up with checking  in all data anyway.
On 7 Jan 2013 16:37, "Jan-willem De Bleser"  wrote:

> On Mon, Jan 7, 2013 at 4:22 PM, Sander Deryckere 
> wrote:
> > Only queries like "give all the houses of street 'Veldstraat'" are better
> > with that binding. But how often do these queries happen?
>
> I haven't found the time yet to give Jo's proposal proper
> consideration, so I'll respond to that and the rest of your mail
> later. I did want to jump in here, however, just to give you another
> example: navigation software. It is given a building in the form of an
> address, and needs this building's associated street so that it can
> find the route along streets from one to another. True, it does need
> to find the closest point on this way, but that is a far smaller
> search than finding the closest matching way out of all nearby data,
> and even easier if there is a driveway.
>
> - Jw
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-be
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 4:22 PM, Sander Deryckere  wrote:
> Only queries like "give all the houses of street 'Veldstraat'" are better
> with that binding. But how often do these queries happen?

I haven't found the time yet to give Jo's proposal proper
consideration, so I'll respond to that and the rest of your mail
later. I did want to jump in here, however, just to give you another
example: navigation software. It is given a building in the form of an
address, and needs this building's associated street so that it can
find the route along streets from one to another. True, it does need
to find the closest point on this way, but that is a far smaller
search than finding the closest matching way out of all nearby data,
and even easier if there is a driveway.

- Jw

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Sander Deryckere]

I completely agree with you, Jo.

Having an easy way to contribute addresses is absolutely necessary. The
walking and cycle nodes can follow complicated relation schemas, because
there are only a few thousand of them, the OSM core users can all map them.
But addresses can't, because everyone needs to help with it.

So being supported by simple tools is a must.

Next to that, I have no problem with redundancy, because it allows to do
certain checks. If names, that should be the same, differ, than there's
clearly something wrong, and someone has to check it locally.

Also, there are two queries that happen most: going from an address as
string to a house object, and going from a house object to an address as
string (known as geocoding and reverse geocoding). Both queries don't need
the binding between a street and house object. Having a streetname tag is
enough to execute them, even if there's no street drawn at all.

Only queries like "give all the houses of street 'Veldstraat'" are better
with that binding. But how often do these queries happen?

So I'm all in favour of adding an addr:street tag to every house, and those
who want it can add an associatedstreet relation with the other tags.

Regards,
Sander
On 7 Jan 2013 12:18, "Jo"  wrote:

> This is what I would propose:
>
> http://www.openstreetmap.org/browse/relation/1919939/history
>
> addr:city, addr:postcode, addr:country, and maybe addr:state, although I'd
> prefer addr:province there go in the associatedStreet relation. The name of
> the street goes into name. This way it becomes usable in the list of
> relations.
>
> http://www.openstreetmap.org/browse/way/141730010/history
>
> addr:housenumber and addr:street are filled out on the buildings and POIs.
>
> The advantage is that this is workable with a minimal editor on a tablet
> or phone, while at the same time reducing tens of thousands of repetitions
> of city names and postcodes.
>
> This way the street name is available as name on the ways and the
> associatedStreet relations and as addr:street on the buildings and the
> POIs. The redundancy that is created this way allows to detect anomalies as
> described by Jan-Willem and get them corrected.
>
> A good example is the problem I just fixed, where a street name of a
> street in Leuven inadvertently got applied to the houses in Reet.
> Geofabrik's site spotted that.
>
> Jo
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-be
>
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

This is what I would propose:

http://www.openstreetmap.org/browse/relation/1919939/history

addr:city, addr:postcode, addr:country, and maybe addr:state, although I'd
prefer addr:province there go in the associatedStreet relation. The name of
the street goes into name. This way it becomes usable in the list of
relations.

http://www.openstreetmap.org/browse/way/141730010/history

addr:housenumber and addr:street are filled out on the buildings and POIs.

The advantage is that this is workable with a minimal editor on a tablet or
phone, while at the same time reducing tens of thousands of repetitions of
city names and postcodes.

This way the street name is available as name on the ways and the
associatedStreet relations and as addr:street on the buildings and the
POIs. The redundancy that is created this way allows to detect anomalies as
described by Jan-Willem and get them corrected.

A good example is the problem I just fixed, where a street name of a street
in Leuven inadvertently got applied to the houses in Reet. Geofabrik's site
spotted that.

Jo
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Marc Gemis]

I've added most of those addresses in Reet.
I use JOSM all the time to create associatedStreet relations.

I vaguely remember seeing a different behavior at a certain point, where
JOSM started adding addr:street tags to the buildings when its building
tool creates an associatedStreet relation. Is this my imagination, or did
anybody else noticed this as well ?

m
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Joren]

Op 07-01-13 11:21, Jan-willem De Bleser schreef:

On Mon, Jan 7, 2013 at 11:14 AM, Jo  wrote:

@Jan-Willem: I wouldn't take away the addr:street from the houses and POIs
when you add them to an associatedStreet relation. I tried that once, a few
years ago, and had to revert because portable editors didn't support
relations (and they still don't) and, when in the field, people are missing
that information then. So it's probably best to keep that redundancy in the
data.

I find that very dangerous, and would rather people added only
addr:street tags than both. What if I incorrectly add an
associatedStreet, and then a mobile mapper adds the correct
addr:street tag - who should the end users then believe? Or if I add
the correct relation, and a vandal a wrong addr:street tag?

You know, seeing as the street already has the name, why is the name
repeated in associatedStreet at all?
I agree for a consistent tagging... Now I don't know what's the best 
street-tag for a building... is a relation better than an addr:street tag?


Joren

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 11:14 AM, Jo  wrote:
> @Jan-Willem: I wouldn't take away the addr:street from the houses and POIs
> when you add them to an associatedStreet relation. I tried that once, a few
> years ago, and had to revert because portable editors didn't support
> relations (and they still don't) and, when in the field, people are missing
> that information then. So it's probably best to keep that redundancy in the
> data.

I find that very dangerous, and would rather people added only
addr:street tags than both. What if I incorrectly add an
associatedStreet, and then a mobile mapper adds the correct
addr:street tag - who should the end users then believe? Or if I add
the correct relation, and a vandal a wrong addr:street tag?

You know, seeing as the street already has the name, why is the name
repeated in associatedStreet at all?

- Jw

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

The remark I made was about the associatedStreet relation. The name of the
street is put in the name tag, just like it is done on the ways forming the
street. So all other tags are addr:country, addr:city, addr:postcode, but
the streetname goes into name.

Everyone is free to add addr:street to the house and POI objects as well.
It's creating an unfortunate redundancy, but as you mention we seem to be
needing that for editors like iLoe and Vespucci which don't support
relations (yet).
Why you need it on Potlatch eludes me, as that one does support relations,
but since I don't have experience with it anymore, I should probably shut
up about it.

I do think associatedStreet relations still have a function to avoid
duplicating/multiplicating the country, city and postcode information that
make addresses complete and possibly to be able to perform quality control
on the data. The redundancy can help to check whether the data is
consistent.

Personally I don't like repeating data over and over, as it makes it easier
to make mistakes and when a streetname changes it becomes necessary to
change it in many places.

@Jan-Willem: I wouldn't take away the addr:street from the houses and POIs
when you add them to an associatedStreet relation. I tried that once, a few
years ago, and had to revert because portable editors didn't support
relations (and they still don't) and, when in the field, people are missing
that information then. So it's probably best to keep that redundancy in the
data.

Jo



2013/1/7 Ivo De Broeck 

> I don't agree with that. Its necessary to have addr:street for every
> address. When you use potlatch on the computer or iLOE on your smartphone
> its easy to bring in new data or correct the data. The associated street is
> redundant (and as i saw in Bierbeek often wrong).
>
> For me it is most important that new users have the possibility of
> introduce new data in a simple way (copy-paste the streetname in
> addr:street). Its a pity that most of the people here give only
> sophisticated solutions for very simple problems.
>
> If you don't make it easy for the contributors, you will never get
> addresses from then.
>
>
> 2013/1/7 Jo 
>
>> The associatedStreet relation has the streetname in 'name', not in
>> addr:street. I also found some relations where this was done incorrectly.
>>
>> It is possible to fix all of them in one go. Advise me if you want me to
>> do so.
>>
>> Polyglot
>>
>>
>> 2013/1/7 Joren 
>>
>>> Op 07-01-13 00:35, Kurt Roeckx schreef:
>>>
>>>  On Sat, Dec 22, 2012 at 03:23:15PM +0100, Sander Deryckere wrote:

> The first thing you notice is that there are a lot of features with
> housenumber information, but without street information. While other
> information (such as city) can be determined from closed boundaries.
> It's
> often ambiguous and hard to determine the street from other OSM
> features.
>
 Osmose counts alot of errors in Belgium because of that.  See:
 http://osmose.openstreetmap.**fr/errors/graph.png?country=**belgium

>>> http://tools.geofabrik.de/**osmi/?view=addresses&lon=4.**
>>> 41356&lat=51.10370&zoom=14&**baselayer=Geofabrik&opacity=1.**
>>> 00&overlays=no_addr_street,**street_not_found
>>>
>>> Geofabrik shows that there are many 'bugs' in the city 'Reet' ... but
>>> when I examine it, some/all houses are tagged with 'associatedStreet
>>> , etc'...
>>> Is this the correct tagging, or do we need to delete that tag, and tag
>>> them with 'addr:street'?
>>>
>>>
>>>
 About 50% of those are because of missing addr:street or
 associatedStreet relation.

 It would in general be a good thing that we try and fix all those
 errors.

  Thanks in advance,
>>> Joren
>>>
>>>
>>> __**_
>>> Talk-be mailing list
>>> Talk-be@openstreetmap.org
>>> http://lists.openstreetmap.**org/listinfo/talk-be
>>>
>>
>>
>> ___
>> Talk-be mailing list
>> Talk-be@openstreetmap.org
>> http://lists.openstreetmap.org/listinfo/talk-be
>>
>>
>
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jan-willem De Bleser]

On Mon, Jan 7, 2013 at 8:17 AM, Jo  wrote:
> The associatedStreet relation has the streetname in 'name', not in
> addr:street. I also found some relations where this was done incorrectly.
>
> It is possible to fix all of them in one go. Advise me if you want me to do
> so.

Do you mean fix associatedStreet relations tagged with addr:street, or
fix buildings tagged with addr:street who are also in an
associatedStreet relation? If the latter, I would not do this
automatically, as any tags that disagree probably need a human to
check them.

On Mon, Jan 7, 2013 at 10:24 AM, Ivo De Broeck  wrote:
> I don't agree with that. Its necessary to have addr:street for every
> address. When you use potlatch on the computer or iLOE on your smartphone
> its easy to bring in new data or correct the data. The associated street is
> redundant (and as i saw in Bierbeek often wrong).
>
> For me it is most important that new users have the possibility of introduce
> new data in a simple way (copy-paste the streetname in addr:street). Its a
> pity that most of the people here give only sophisticated solutions for very
> simple problems.

Unfortunately, simple solutions often only solve some of the problems.
associatedStreet is more complex than addr:street, but it makes clear
precisely to which street a building belongs. Otherwise, if you need
the street that belongs to that building, you have to search for the
geographically closest way with the same name, a much slower, more
time-consuming and not necessarily accurate operation.

I don't object to either being used at this time, however. I add
associatedStreet relations when there is *no* address information but
I don't convert addr:street addresses to new relations. I do fix it,
one way or the other, when I see them used simultaneously on the same
building, because that is indeed redundant.

Cheers,
Jw

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Ivo De Broeck]

I don't agree with that. Its necessary to have addr:street for every
address. When you use potlatch on the computer or iLOE on your smartphone
its easy to bring in new data or correct the data. The associated street is
redundant (and as i saw in Bierbeek often wrong).

For me it is most important that new users have the possibility of
introduce new data in a simple way (copy-paste the streetname in
addr:street). Its a pity that most of the people here give only
sophisticated solutions for very simple problems.

If you don't make it easy for the contributors, you will never get
addresses from then.

2013/1/7 Jo 

> The associatedStreet relation has the streetname in 'name', not in
> addr:street. I also found some relations where this was done incorrectly.
>
> It is possible to fix all of them in one go. Advise me if you want me to
> do so.
>
> Polyglot
>
>
> 2013/1/7 Joren 
>
>> Op 07-01-13 00:35, Kurt Roeckx schreef:
>>
>>  On Sat, Dec 22, 2012 at 03:23:15PM +0100, Sander Deryckere wrote:
>>>
 The first thing you notice is that there are a lot of features with
 housenumber information, but without street information. While other
 information (such as city) can be determined from closed boundaries.
 It's
 often ambiguous and hard to determine the street from other OSM
 features.

>>> Osmose counts alot of errors in Belgium because of that.  See:
>>> http://osmose.openstreetmap.**fr/errors/graph.png?country=**belgium
>>>
>> http://tools.geofabrik.de/**osmi/?view=addresses&lon=4.**
>> 41356&lat=51.10370&zoom=14&**baselayer=Geofabrik&opacity=1.**
>> 00&overlays=no_addr_street,**street_not_found
>>
>> Geofabrik shows that there are many 'bugs' in the city 'Reet' ... but
>> when I examine it, some/all houses are tagged with 'associatedStreet
>> , etc'...
>> Is this the correct tagging, or do we need to delete that tag, and tag
>> them with 'addr:street'?
>>
>>
>>
>>> About 50% of those are because of missing addr:street or
>>> associatedStreet relation.
>>>
>>> It would in general be a good thing that we try and fix all those
>>> errors.
>>>
>>>  Thanks in advance,
>> Joren
>>
>>
>> __**_
>> Talk-be mailing list
>> Talk-be@openstreetmap.org
>> http://lists.openstreetmap.**org/listinfo/talk-be
>>
>
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-be
>
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

The associatedStreet relation has the streetname in 'name', not in
addr:street. I also found some relations where this was done incorrectly.

It is possible to fix all of them in one go. Advise me if you want me to do
so.

Polyglot

2013/1/7 Joren 

> Op 07-01-13 00:35, Kurt Roeckx schreef:
>
>  On Sat, Dec 22, 2012 at 03:23:15PM +0100, Sander Deryckere wrote:
>>
>>> The first thing you notice is that there are a lot of features with
>>> housenumber information, but without street information. While other
>>> information (such as city) can be determined from closed boundaries. It's
>>> often ambiguous and hard to determine the street from other OSM features.
>>>
>> Osmose counts alot of errors in Belgium because of that.  See:
>> http://osmose.openstreetmap.**fr/errors/graph.png?country=**belgium
>>
> http://tools.geofabrik.de/**osmi/?view=addresses&lon=4.**
> 41356&lat=51.10370&zoom=14&**baselayer=Geofabrik&opacity=1.**
> 00&overlays=no_addr_street,**street_not_found
>
> Geofabrik shows that there are many 'bugs' in the city 'Reet' ... but when
> I examine it, some/all houses are tagged with 'associatedStreet
> , etc'...
> Is this the correct tagging, or do we need to delete that tag, and tag
> them with 'addr:street'?
>
>
>
>> About 50% of those are because of missing addr:street or
>> associatedStreet relation.
>>
>> It would in general be a good thing that we try and fix all those
>> errors.
>>
>>  Thanks in advance,
> Joren
>
>
> __**_
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.**org/listinfo/talk-be
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Joren]

Op 07-01-13 00:35, Kurt Roeckx schreef:

On Sat, Dec 22, 2012 at 03:23:15PM +0100, Sander Deryckere wrote:

The first thing you notice is that there are a lot of features with
housenumber information, but without street information. While other
information (such as city) can be determined from closed boundaries. It's
often ambiguous and hard to determine the street from other OSM features.

Osmose counts alot of errors in Belgium because of that.  See:
http://osmose.openstreetmap.fr/errors/graph.png?country=belgium

http://tools.geofabrik.de/osmi/?view=addresses&lon=4.41356&lat=51.10370&zoom=14&baselayer=Geofabrik&opacity=1.00&overlays=no_addr_street,street_not_found

Geofabrik shows that there are many 'bugs' in the city 'Reet' ... but 
when I examine it, some/all houses are tagged with 'associatedStreet 
, etc'...
Is this the correct tagging, or do we need to delete that tag, and tag 
them with 'addr:street'?




About 50% of those are because of missing addr:street or
associatedStreet relation.

It would in general be a good thing that we try and fix all those
errors.


Thanks in advance,
Joren

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Kurt Roeckx]

On Sat, Dec 22, 2012 at 03:23:15PM +0100, Sander Deryckere wrote:
> 
> The first thing you notice is that there are a lot of features with
> housenumber information, but without street information. While other
> information (such as city) can be determined from closed boundaries. It's
> often ambiguous and hard to determine the street from other OSM features.

Osmose counts alot of errors in Belgium because of that.  See:
http://osmose.openstreetmap.fr/errors/graph.png?country=belgium

About 50% of those are because of missing addr:street or
associatedStreet relation.

It would in general be a good thing that we try and fix all those
errors.


Kurt


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

Frederik Ramm developed a script that takes history into account:

http://www.remote.org/frederik/tmp/housenumbers.html

It does a lot more credit to the work of Ivodeb around Korbeek-Lo and
Bierbeek. The numbers are also for worldwide edits instead of only Belgium.

Happy New Year (well almost),

Jo
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

2012/12/24 A.Pirard.Papou 

>  On 2012-12-23 22:21, Jo wrote :
>
> Hi Chris,
>
> The easiest way is to simply add addr:housenumber nodes. It's hard to do
> anything wrong with that.
>
> If the building is an area, it's a good idea to put the number on a node
> instead of on the way, so that it doesn't fight with the restaurant name
> for room inside the area.  Plenty of examples around 
> here.
> But you may prefer not to tag for the renderer ;-)
>
> One step further is to add addr:interpolation vectors. Add the numbers on
> the streetcorners and the special cases like 5A (also add 5 and 6 in that
> case) and a way in between.
>
> This recalls me when I built my house.  The numbers next were 1 and 3.
> Went to the adm and explained that.
> - no problem, you'll get n°2
> - (surprised that n°2 isn't opposite) OK, but there are two houses between
> 1 and 3
> - no problem, you'll get n° 2A, see in the box, 2A and 2B are prepared
> - OK, but my house is the second one
> - no problem, you get n° 2B, then
> No problem, OSM did not exist yet, because it lasted a few years until all
> the house were renumbered and the place name changed, some persons were in
> another street too.
>

Normally one would expect 5A, 5B, etc to be between 5 and 7, but I have
found streets where 1A and 1B come before 1. It all depends on which houses
were built first, I presume.

Jo
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[A.Pirard.Papou]

On 2012-12-23 22:21, Jo wrote :

Hi Chris,

The easiest way is to simply add addr:housenumber nodes. It's hard to 
do anything wrong with that.
If the building is an area, it's a good idea to put the number on a node 
instead of on the way, so that it doesn't fight with the restaurant name 
for room inside the area. Plenty of examples around here 
.  But you may prefer 
not to tag for the renderer ;-)
One step further is to add addr:interpolation vectors. Add the numbers 
on the streetcorners and the special cases like 5A (also add 5 and 6 
in that case) and a way in between.

This recalls me when I built my house.  The numbers next were 1 and 3.
Went to the adm and explained that.
- no problem, you'll get n°2
- (surprised that n°2 isn't opposite) OK, but there are two houses 
between 1 and 3

- no problem, you'll get n° 2A, see in the box, 2A and 2B are prepared
- OK, but my house is the second one
- no problem, you get n° 2B, then
No problem, OSM did not exist yet, because it lasted a few years until 
all the house were renumbered and the place name changed, some persons 
were in another street too.


Cheers,

André.


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Sander Deryckere]

Op 23 dec. 2012 21:00 schreef "Chris Van Bael" 
het volgende:
>
> On 23-dec.-2012, at 12:28, Sander Deryckere  wrote:
>
>> Now the street network is good as complete (we even have streets in the
DB that are still under construction). So we need to focus on other things.
Those things are (according to me):
>>
>>
>> Addresses including housenumbers
>> POI's including extra information s.a. opening hours
>> Extra info on roads (maxspeed, lanes ...)
>
>
> I've been lurking on this list for too long, but this message triggered
me.
> A few years ago I added a few streets in my neighbourhood, but indeed now
all the streets are in OSM, so more detailed info needs to be added.
> Problem is: I don't know how to do this in a decent way!
> And I'm very afraid to break something, or even worse do it in a bad way.
> There have been too many discussions on this list about bad conventions
used by others to give me confidence to add these things.

When you are adding stuff that can't be guessed from the area images, it's
hard to do something wrong. And when it's wrong, it's always easy to
correct.

So just do what you think is good, and if nobody tells you it is wrong,
that means you're doing great.

As long as you stay away from armchair mapping and batch editing, you will
never have to worry about these discussions (unless you really want ).

> What would help me (and others I believe), is that there are simple
tutorials to execute these things that clearly describe the conventions.

Maybe that's an idea, but a lot of users have their own preferred ways of
mapping, based on their favorite tools and available hardware. It's
difficult to find a standard in it.
>
> For the housenumbers for example : is there an easy way to add these
through my Android phone?
> Or how do you do this?
>
How I do it? If I am prepared, I have a printed map (at high zoom level,
via the walkingpapers service ) of the area I want to map, then I make
notes on it, and process them afterwards.

When I'm not prepared, I just use OSMtracker and make pictures of the
housenumbers and other features, when I come home, JOSM shows me where
these pictures were taken, and from the memory (and picture background ) I
know which house it is (left or right? First or second?).

I don't really like inputting data directly from my phone, but maybe that's
because I have an old, slow Android with a small screen.

> Thanks,
>
> Chris
>

>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[A.Pirard.Papou]

  
  
On 2012-12-22 15:23, Sander Deryckere
  wrote :

...
  Now, how many addresses would be missing. We can't assume Belgium
  has 11 million addresses, as many people live together. So I
  searched other data. The number of addresses in Belgium seems
  impossible to find, but I did find the number of families in
  Belgium: http://www.centrumvoorsociaalbeleid.be/indicatoren/index.php?q=node/176.
  I assume that the number of addresses must be about the same.
  There are addresses without families (like firms) and multiple
  families living in one apartment with one address (but often
  different post boxes). 

 (of
  course) is maintaining a database of the number of mailboxes per
  distribution area, so that the people doing mass mailing, like
  printers, know how many mail items to provide. 
It's free, but you need to subscribe here to access the service.
http://www.bpost.be/distripost/outil_de_reservation.html
http://www.bpost.be/distripost/reservatietool.html
This subscription, including receiving snail mail and a password,
looks like the business status of the requester is verified.  I'll
send by private e-mail to anyone requesting it within next week a
more descriptive French text written by a user for his customers.  I
could have a peek at the data and I can describe it as OSM valuable.

Also,
  I came across this file.  I couldn't find the same for other
municipalities, but it's format makes me think that it's data that
each administration must or should maintain and that it could be
available on request.  I could blast the PDF and get each street on
one line of t.txt.
perl -ne 'print if s/(\d+)\/(\d+) (.*?)\s+(\d+)$/$1,$2,$3,$4/' t.txt
> t.asc
did the rest and I pasted t.asc in a spreadsheet.
I made a additional column to contain the village/hamlet name and a
simple SUMIF per area is adding up the persons count for each of
them.

I welcome under this subject any feedback regarding both sources.

Given mailboxes and population counts, you might find that their
ratio is fairly constant, maybe after separating mainly urban vs
rural statistics.

How to map?
  
  There's also a lot of armchair mapping that can be done. First of
  all, all that streetnames that need to be added. People are better
  in guessing the right streetname, and if there's doubt, just add a
  fixme tag.

I often do that and I had just did it the day before you wrote
this.  www.restaurant indicated 6, street X and when I curiously
checked with GoogleMaps: it did not know the restaurant, located 6,
street X very far away from there and called street X street Y.
Conclusion: beware: some people don't know where they are living!!!
;-)
Next to that, if you see a restaurant on the map,
  without address data, just search the website of that restaurant
  and get the address data from there. You're doing nothing wrong,
  as long as you don't take the data from a database (such as the
  golden pages), you aren't violating any copyrights or database
  rights. While you're at the website of the restaurant, you can
  also add other information s.a. opening hours or phone number.

The scrupulous mind can get the number from the YP and phone the
restaurant to ask them their phone number.  Or you can learn the YP
by heart, and do the mapping two days after.  Or ask your wife to
look it up and phone it to the neighbor; he will probably phone back
and ask what's going on there.  Lots of ©less methods ;-)

I usually refrain from indicating phone numbers. There are already
so many problems with web addresses changes that could be easily
detected but are not cared for, that it's useless to add phone
number changes that cannot be detected.  I would give priority to
pop up a tip when the mouse hovers over the restaurant with a Web
link to click to see the rest.  I drew the attention of JOSM preset
writers that a Web address is often much wider that what's usually
called a Contact.

Happy Xmas, be it in street X or Y.


  

  André.

  



  

  

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

Hi Chris,

The easiest way is to simply add addr:housenumber nodes. It's hard to do
anything wrong with that. For good measure add addr:street as well. On
Android you can use the latest version of Vespucci. It works quite well,
but it's not a lot of use without an internet connection (3G or WLAN).

One step further is to add addr:interpolation vectors. Add the numbers on
the streetcorners and the special cases like 5A (also add 5 and 6 in that
case) and a way in between.

I prefer to add addr:housenumber in 'the field' and do 'postprocessing'
afterwards. Drawing the buildings and adding them to associatedStreet
relations, but the most important part are those addr:housenumber nodes.
The rest can be done later, but that's the 'raw' data.

Jo

2012/12/23 Chris Van Bael 

> On 23-dec.-2012, at 12:28, Sander Deryckere  wrote:
>
> Now the street network is good as complete (we even have streets in the DB
> that are still under construction). So we need to focus on other things.
> Those things are (according to me):
>
>
>
>- Addresses including housenumbers
>- POI's including extra information s.a. opening hours
>- Extra info on roads (maxspeed, lanes ...)
>
>
> I've been lurking on this list for too long, but this message triggered me.
> A few years ago I added a few streets in my neighbourhood, but indeed now
> all the streets are in OSM, so more detailed info needs to be added.
> Problem is: I don't know how to do this in a decent way!
> And I'm very afraid to break something, or even worse do it in a bad way.
> There have been too many discussions on this list about bad conventions
> used by others to give me confidence to add these things.
> What would help me (and others I believe), is that there are simple
> tutorials to execute these things that clearly describe the conventions.
>
> For the housenumbers for example : is there an easy way to add these
> through my Android phone?
> Or how do you do this?
>
> Thanks,
>
> Chris
>
>
> ___
> Talk-be mailing list
> Talk-be@openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-be
>
>
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Chris Van Bael]

On 23-dec.-2012, at 12:28, Sander Deryckere  wrote:

> Now the street network is good as complete (we even have streets in the DB 
> that are still under construction). So we need to focus on other things. 
> Those things are (according to me):
> 
> Addresses including housenumbers
> POI's including extra information s.a. opening hours
> Extra info on roads (maxspeed, lanes ...)

I've been lurking on this list for too long, but this message triggered me.
A few years ago I added a few streets in my neighbourhood, but indeed now all 
the streets are in OSM, so more detailed info needs to be added.
Problem is: I don't know how to do this in a decent way!
And I'm very afraid to break something, or even worse do it in a bad way.
There have been too many discussions on this list about bad conventions used by 
others to give me confidence to add these things.
What would help me (and others I believe), is that there are simple tutorials 
to execute these things that clearly describe the conventions.

For the housenumbers for example : is there an easy way to add these through my 
Android phone?
Or how do you do this?

Thanks,

Chris

___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Glenn Plas]

PS: I'm using a lot of geocoding when I truly mean 'reverse
geocode',   I only focus in getting full addresses back from
coordinates.


That's probably the main difference. I focus on geocoding (getting 
coordinates from an address). And when you have long streets without 
housenumbers, Nominatim will give a result, but the result can easily 
be 20 km off.  Which is really unacceptable. As you don't know how 
long the street is, you can only assume that you have the correct 
place when the housenumber is found.
20 Km is unacceptable, but It's probably partly the fault of nominatim 
code itself, I've been doing plenty of mods and the way it finds 
(geocode) places is pretty complicated.  I'm not saying incorrect but it 
sure is huge (try forking it on github to find out).   And it's based on 
getting USA/GB address types in general. Which means, when we deviate a 
bit from the 'current' de facto standard that nominatim uses, we get 
less accurate results.   That doesn't mean it's not in OSM data.


I've been playing a lot with the idea of creating a light-weight 
geocoder api on top of Gazetteer.   It also always intruiged me why I 
was unable to get a postcode (but I did get a city) in a reverse geocode 
on my own servers (at one point I installed enough gazetteers to know 
the procedure by heart).  But the public one did find postal codes.  I 
always used all their docs for my own installs.  That triggered me to 
patch the result set and include postal codes by name and distance 
(coordinates as a source helps). Which turned problematic in 
Brussels/Antwerp sometimes.   But the customers didn't really care if it 
was 1 code off.


So that's what I meant with the "1 in 40" statement.
In that sense, you are probably right.  You need to know how to feed 
keywords in the correct order in the search page.   In google you can 
pretty much slam anything in it whatever the order, but they probably 
use a 1000 of their nodes at once to figure that out.




And I agree that the data is becoming better and better. When I 
started, my own street wasn't on the map (this triggered me). Now the 
street network is good as complete (we even have streets in the DB 
that are still under construction). So we need to focus on other 
things. Those things are (according to me):


I love to add those under constructions, I check the site of my 
municipality frequently for their names etc.   I love to remap places 
that have recently been reconstructed, like the R6 in 
Mechelen/Sint-Katelijne-Waver for example.  It took a year for google to 
catch up.  That's how I push people into mapping or make customers aware 
that their name is on that map because I put it there.  They absolutely 
love that stuff.




  * Addresses including housenumbers
  * POI's including extra information s.a. opening hours
  * Extra info on roads (maxspeed, lanes ...)


agreed with all.   The easiest way I find to add addresses/street info is;
 - know the place
 - put all merchants in it ('zelfstandigen', bakery, chinese food, 
vending machines).   Use their site or the municipality to find the 
addresses, every place does that with the small shops.   They won't mind 
this at all, in fact, they love the exposure.  And it's an official list 
and you can use it, it is public information by nature.
 - Just start with what you know.  Try finding bread here in the area 
at night, only for that reason I add all those vending machines so I can 
follow the android in my quest for food ;-)
I just wanted to see how our addresses are evolving, as a result of 
the other blog posts.and it wasn't that hard to get some indicative 
numbers.
It's useful exercise, which I applaud.  Only at the very end of my 
initial reaction I started to realise the 'reverse geocoding' fact isn't 
typical use for everyone ;-)


In that sense, there is a german oriented plugin to add maxspeed/road 
signs to JOSM. It's a great tool and a way to put some standards in.  It 
has a way to customize, I did like 5% of it to map road signs.  It 
supports mapping it on a way and just signs next to the road( less 
useful imho).   'road signs plugin' it's called.   So I'm trying to make 
a Belgian version, in the recent road signs discussion this could prove 
useful for extra quality.


Mvg,

Glenn
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Sander Deryckere]

2012/12/23 Glenn Plas 

> Sander/Jo
>
> Lovely stats, but I don't really agree with all the conclusions, plenty of
> doubtful assumptions as well stated, but I will go along with those for a
> second.
>
> I work professionally with OSM data, more specifically Nominatim for years
> now(2007/8), If you do not account for the house numbers, the results are
> far better than google(is)/yahoo(was).
>
> The engines that perform the best reverse geocoding results in order :
>  nominatim, google, yahoo, bing, ...rest
>
> In reality the only thing that is preventing a full geocode (reverse)
> result is the missing housenumber.  The streets geocode fine, with better
> quality than the number 1  (google at this moment).  Professionally I don't
> care too much for house numbers (albeit I understand the need for this in
> other applications of course).  All commercial ones have easter eggs
> everywhere, I found more than 20 this year in google, streets that don't
> exist so they can see if they get copied over.  These will be returned in
> geocode results.
>
> So forgive me, but reading that I will only get 1 hit out of 40 is just
> not correct compared to reality.  I am able to geocode everything I throw
> at  nominatim,  I'm not talking about 1000 geocodes a week but millions (5
> to 10).   Only 5 % is being relayed to a different engine since the DB I
> talk about is a Benelux OSM dump, not a full, so anything outside the
> benelux will be sent to external engines.
>
> I would accept those numbers if you would state: "Getting back a full
> address including housenumbers will probably work 1 out of 40" But you WILL
> get an address back from Nominatim in all other cases, without a house
> number of course.
>
> It surprises me a lot that I'm even in those top stats with a mere 111
> addresses added I don't think we can talk about a success in .BE
> looking at those totals.   But I do agree its improving, compared to 2008
> nominatim result sets.  it not only improved in source quality (OSM) but
> also implementation (so postal code is returned now, it didn't use to do
> that).
>
> Happy holidays!
>
> PS: I'm using a lot of geocoding when I truly mean 'reverse geocode',   I
> only focus in getting full addresses back from coordinates.
>

That's probably the main difference. I focus on geocoding (getting
coordinates from an address). And when you have long streets without
housenumbers, Nominatim will give a result, but the result can easily be 20
km off.  Which is really unacceptable. As you don't know how long the
street is, you can only assume that you have the correct place when the
housenumber is found.

So that's what I meant with the "1 in 40" statement.

And I agree that the data is becoming better and better. When I started, my
own street wasn't on the map (this triggered me). Now the street network is
good as complete (we even have streets in the DB that are still under
construction). So we need to focus on other things. Those things are
(according to me):


   - Addresses including housenumbers
   - POI's including extra information s.a. opening hours
   - Extra info on roads (maxspeed, lanes ...)

I just wanted to see how our addresses are evolving, as a result of the
other blog posts.and it wasn't that hard to get some indicative numbers.

Regards,
Sander
___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Glenn Plas]

Sander/Jo

Lovely stats, but I don't really agree with all the conclusions, plenty 
of doubtful assumptions as well stated, but I will go along with those 
for a second.


I work professionally with OSM data, more specifically Nominatim for 
years now(2007/8), If you do not account for the house numbers, the 
results are far better than google(is)/yahoo(was).


The engines that perform the best reverse geocoding results in order :  
nominatim, google, yahoo, bing, ...rest


In reality the only thing that is preventing a full geocode (reverse) 
result is the missing housenumber.  The streets geocode fine, with 
better quality than the number 1  (google at this moment).  
Professionally I don't care too much for house numbers (albeit I 
understand the need for this in other applications of course).  All 
commercial ones have easter eggs everywhere, I found more than 20 this 
year in google, streets that don't exist so they can see if they get 
copied over.  These will be returned in geocode results.


So forgive me, but reading that I will only get 1 hit out of 40 is just 
not correct compared to reality.  I am able to geocode everything I 
throw at  nominatim,  I'm not talking about 1000 geocodes a week but 
millions (5 to 10).   Only 5 % is being relayed to a different engine 
since the DB I talk about is a Benelux OSM dump, not a full, so anything 
outside the benelux will be sent to external engines.


I would accept those numbers if you would state: "Getting back a full 
address including housenumbers will probably work 1 out of 40" But you 
WILL get an address back from Nominatim in all other cases, without a 
house number of course.


It surprises me a lot that I'm even in those top stats with a mere 111 
addresses added I don't think we can talk about a success in .BE 
looking at those totals.   But I do agree its improving, compared to 
2008 nominatim result sets.  it not only improved in source quality 
(OSM) but also implementation (so postal code is returned now, it didn't 
use to do that).


Happy holidays!

PS: I'm using a lot of geocoding when I truly mean 'reverse geocode',   
I only focus in getting full addresses back from coordinates.


Glenn


___
Talk-be mailing list
Talk-be@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk-be

Re: [OSM-talk-be] Addresses in Belgium

[Jo]

2012/12/22 Sander Deryckere 

> After the articles about  the addresses in Germany (where every average
> German contributor should gather 1000 addresses), I wondered how we were
> doing in Belgium. So I downloaded the address information in Belgium and
> did some counting.
>
> I only counted the addr:housenumber and addr:street tags. I didn't occupy
> myself with the associatedstreet relations, as I've not that often seen
> those on their own. If you think I'm wrong, I can do my counting again. I
> also didn't take the addr:interpolation into account. Just because it's
> difficult to analyse it.
>
>
>
> The first thing you notice is that there are a lot of features with
> housenumber information, but without street information. While other
> information (such as city) can be determined from closed boundaries. It's
> often ambiguous and hard to determine the street from other OSM features.
>
> Some applications (s.a. Nominatim) implement the "street guessing",
> sometimes with wrong results. Other apps (s.a. OsmAnd) just don't include
> houses without street information in their search, so that info is
> completely lost.
>
> Now, how many addresses would be missing. We can't assume Belgium has 11
> million addresses, as many people live together. So I searched other data.
> The number of addresses in Belgium seems impossible to find, but I did find
> the number of families in Belgium:
> http://www.centrumvoorsociaalbeleid.be/indicatoren/index.php?q=node/176.
> I assume that the number of addresses must be about the same. There are
> addresses without families (like firms) and multiple families living in one
> apartment with one address (but often different post boxes).
>
> So that means we're needing about 4.5 million addresses and currently have
> 112 000. The completeness is thus about 2.5% of address data. Not a very
> good number. When searching 40 addresses, only 1 on average will be found
> in OSM.
>
> But it becomes better when we look how the data evolved. Of that 112 000
> addresses, there are 76 000 created (or modified) in 2012 and 105 000 since
> 2011. That means that the number of addresses created is going up  If we
> can keep a bit of growth, we could map the majority of addresses in a few
> years.
>
> So continue with the effort, and map as many addresses as possible.
>
> *How to map?*
>
> There's also a lot of armchair mapping that can be done. First of all, all
> that streetnames that need to be added. People are better in guessing the
> right streetname, and if there's doubt, just add a fixme tag.
>
> Next to that, if you see a restaurant on the map, without address data,
> just search the website of that restaurant and get the address data from
> there. You're doing nothing wrong, as long as you don't take the data from
> a database (such as the golden pages), you aren't violating any copyrights
> or database rights. While you're at the website of the restaurant, you can
> also add other information s.a. opening hours or phone number.
>
> Of course, when the weather is good, you can go out and map addresses. I
> normally use photo mapping because it's so fast (and if you see an other
> feature, you can also just take a picture of it), but there are also apps
> for that, s.a. the Keypadmapper app for Android.
>

I was curious about some actual statistics, so I used Sander's Overpass
query (somewhat modified to include asscociatedStreet relations):

area[name="België - Belgique - Belgien"];
> (
>   node(area);
>   <;
> ) -> .allnodeswaysrelationsinBelgium;
> (
>   rel.allnodeswaysrelationsinBelgium["type"="associatedStreet"];
> ) -> .allassociatedStreetrelations;
> (
>   rel.allnodeswaysrelationsinBelgium["addr:housenumber"];
> ) -> .alladdr_housenumberrelations;
> (
>   way.allnodeswaysrelationsinBelgium["addr:housenumber"];
> ) -> .alladdr_housenumberways;
> (
>   node.allnodeswaysrelationsinBelgium["addr:housenumber"];
> ) -> .alladdr_housenumbernodes;
> (
>   .allassociatedStreetrelations;
>   .allassociatedStreetrelations >;
>   .alladdr_housenumberrelations;
>   .alladdr_housenumberrelations >;
>   .alladdr_housenumberways;
>   .alladdr_housenumberways >;
>   .alladdr_housenumbernodes;
> );
> out meta qt;
>

Then some Python magic with SAX to extract the users from the xml:

import xml.sax
>
> ''' sample data:
>timestamp="2011-04-17T09:51:55Z" changeset="7884790" uid="9176"
> user="Maarten Deen">
> 
> 
> 
> 
> 
> 
> 
> 
>   
>changeset="11710662" uid="436365" user="escada">
> 
> 
> 
> 
> 
> 
> 
> 
> 
>   
> '''
>
> class OSMContentHandler(xml.sax.ContentHandler):
> def __init__(self):
> xml.sax.ContentHandler.__init__(self)
> self.noteattributes = {}
> self.tags = {}
> self.users = {}
>
> def startElement(self, tagname, attrs):
> if tagname == "tag":
> self.tags[attrs.getValue("k")] = attrs.getValue("v")
> elif tagname in ["node", "way", "re