Re: [Talk-ca] Mapping buildings in Canada by 2020

[Tim Elrick]

There are in fact, quite some German based mappers, mapping in Canada.
So, you are right: let's just get more active mappers - worldwide; it
is, however, still easier to activate them locally.

I will pass on your positive experience with JOSM to my group for the
next event.

Cheers, Tim

Am 23.11.2017 um 10:32 schrieb john whelan:
>>I would have to run a query now to find out if the relative number of
active mappers is higher in one country than the other, but that's not
my point.

But how do you determine where a mapper lives and don't forget many
armchair mappers map in a different location to where they live.

JOSM and the building_tool plugin worked very well for our lot.  We did
ask them to come with JAVA installed and we only taught them enough JOSM
to map a building with the plugin.  Well we showed them a few more
things as they got more comfortable with it.

Cheerio John

On 23 November 2017 at 10:26, Tim Elrick > wrote:

Hi John,

Thanks for your feedback and background information.

I think, we are on the same page. I am concerned with quality too, while
mapping should remain enjoyable.

We shied away from JOSM for newbies because it seemed more technical to
my groups members. I personally like JOSM better, and the building
plug-in is great. Maybe I manage to convince the group to use it
next time.

I did not intend to call for experienced mappers to do all the
validation (I know it is tedious; however, correcting and esp. updating
makes OSM great and in some place much better than the official
sources). I think, that the group who initiated the mapping should
'clean up after themselves' (and I just wanted to affirm that we will do
that). I just wanted to express gratitude to mappers how do help out.

Once I am more into it, I am happy to help out validating other's work.

I did not mean to cheery pick when I quoted the validation website (I
very much appreciate the wiki page). I just wanted to make a point about
timing.

Regarding Canada, as a geographer I am fully aware of the fact Canada
having relatively less population, however, it has still almost half of
the population of Germany and the urban areas, which most of OSM mappers
are concerned with, might be relatively (to population) similar in size
(that's just a guess). I would have to run a query now to find out if
the relative number of active mappers is higher in one country than the
other, but that's not my point. The relative numbers do not matter, as
actual people do the mapping. And there, I hope we agree, the Canadian
OSM community could do with more active mappers.

Tim


Am 23.11.2017 um 07:54 schrieb john whelan:
The issue is the quality of the mapping, nothing else. I attended one of
these geoweek events and we used JOSM with the building_tool plugin. 
The mapping of buildings was accurate even though 75% of the mappers had
never used JOSM before.  There was no formal validation done but I
verified each mappers work as they did it.  I got the impression that
the mappers enjoyed the exercise and I think for me that was the most
important thing.  Mapping should be fun.

There was no mention of the work would be validated nor did we record
the mappers userids to ensure which mappers had mapped.  Other mappers
had marked tiles done on the grid.

I was under the impression that Stats Canada was involved but was later
assured by them that this was not the case.

The problem of lots of new mappers producing low quality work really
reared its head during the Nepal crisis.  I do mainly validation on HOT
projects in Africa and I ended up pulling in chunks of Africa and just
trying to clean up the map.  Currently I'm looking at one mapper who has
added more than a thousand ways with one tag I think it says
source=PGS.  Data quality is a major issue in OpenStreetMap.  Recently
someone gave up when looking for area=yes or buildings drawn in iD but
left untagged for the most part.  I think in Europe it was 100,000 or
more worldwide it was far higher and that's when the person looking at
it gave up.

There are many examples in Africa of groups of buildings being mapped as
one building and labelled building=house.  That's what we are trying to
avoid.  It is possible to correctly map a building in iD I've seen it
done but it takes time.  It is far easier to sort of roughly get it
right and roughly means not accurately.  I think the thing we need to
avoid is a feeling the mapper needs to get a tile done. That's when they
start to rush things.

Building validation?  I can think of no validator who enjoys having to
take two or three times longer to correct someones's work than it would
take them to map it in JOSM with the building_tool in the first place. 
I'm 

Re: [Talk-ca] Mapping buildings in Canada by 2020

[john whelan]

>I would have to run a query now to find out if the relative number of
active mappers is higher in one country than the other, but that's not my
point.

But how do you determine where a mapper lives and don't forget many
armchair mappers map in a different location to where they live.

JOSM and the building_tool plugin worked very well for our lot.  We did ask
them to come with JAVA installed and we only taught them enough JOSM to map
a building with the plugin.  Well we showed them a few more things as they
got more comfortable with it.

Cheerio John

On 23 November 2017 at 10:26, Tim Elrick  wrote:

> Hi John,
>
> Thanks for your feedback and background information.
>
> I think, we are on the same page. I am concerned with quality too, while
> mapping should remain enjoyable.
>
> We shied away from JOSM for newbies because it seemed more technical to
> my groups members. I personally like JOSM better, and the building
> plug-in is great. Maybe I manage to convince the group to use it next time.
>
> I did not intend to call for experienced mappers to do all the
> validation (I know it is tedious; however, correcting and esp. updating
> makes OSM great and in some place much better than the official
> sources). I think, that the group who initiated the mapping should
> 'clean up after themselves' (and I just wanted to affirm that we will do
> that). I just wanted to express gratitude to mappers how do help out.
>
> Once I am more into it, I am happy to help out validating other's work.
>
> I did not mean to cheery pick when I quoted the validation website (I
> very much appreciate the wiki page). I just wanted to make a point about
> timing.
>
> Regarding Canada, as a geographer I am fully aware of the fact Canada
> having relatively less population, however, it has still almost half of
> the population of Germany and the urban areas, which most of OSM mappers
> are concerned with, might be relatively (to population) similar in size
> (that's just a guess). I would have to run a query now to find out if
> the relative number of active mappers is higher in one country than the
> other, but that's not my point. The relative numbers do not matter, as
> actual people do the mapping. And there, I hope we agree, the Canadian
> OSM community could do with more active mappers.
>
> Tim
>
>
> Am 23.11.2017 um 07:54 schrieb john whelan:
> The issue is the quality of the mapping, nothing else. I attended one of
> these geoweek events and we used JOSM with the building_tool plugin.
> The mapping of buildings was accurate even though 75% of the mappers had
> never used JOSM before.  There was no formal validation done but I
> verified each mappers work as they did it.  I got the impression that
> the mappers enjoyed the exercise and I think for me that was the most
> important thing.  Mapping should be fun.
>
> There was no mention of the work would be validated nor did we record
> the mappers userids to ensure which mappers had mapped.  Other mappers
> had marked tiles done on the grid.
>
> I was under the impression that Stats Canada was involved but was later
> assured by them that this was not the case.
>
> The problem of lots of new mappers producing low quality work really
> reared its head during the Nepal crisis.  I do mainly validation on HOT
> projects in Africa and I ended up pulling in chunks of Africa and just
> trying to clean up the map.  Currently I'm looking at one mapper who has
> added more than a thousand ways with one tag I think it says
> source=PGS.  Data quality is a major issue in OpenStreetMap.  Recently
> someone gave up when looking for area=yes or buildings drawn in iD but
> left untagged for the most part.  I think in Europe it was 100,000 or
> more worldwide it was far higher and that's when the person looking at
> it gave up.
>
> There are many examples in Africa of groups of buildings being mapped as
> one building and labelled building=house.  That's what we are trying to
> avoid.  It is possible to correctly map a building in iD I've seen it
> done but it takes time.  It is far easier to sort of roughly get it
> right and roughly means not accurately.  I think the thing we need to
> avoid is a feeling the mapper needs to get a tile done. That's when they
> start to rush things.
>
> Building validation?  I can think of no validator who enjoys having to
> take two or three times longer to correct someones's work than it would
> take them to map it in JOSM with the building_tool in the first place.
> I'm unable to even think of a case where a project has been validated
> and the buildings corrected.  When I validate I'm trying to correct the
> mapper's work and give them feedback so they will map more accurately in
> future.  There is no point in doing this to someone who will map once.
> It's a waste of my time.
>
> The wiki page you pointed to, I wrote much of it. the most important
> part which you skipped is feedback from a user.
>
>
>   Why do we validate?
>
> 

Re: [Talk-ca] Mapping buildings in Canada by 2020

[Tim Elrick]

Hi John,

Thanks for your feedback and background information.

I think, we are on the same page. I am concerned with quality too, while
mapping should remain enjoyable.

We shied away from JOSM for newbies because it seemed more technical to
my groups members. I personally like JOSM better, and the building
plug-in is great. Maybe I manage to convince the group to use it next time.

I did not intend to call for experienced mappers to do all the
validation (I know it is tedious; however, correcting and esp. updating
makes OSM great and in some place much better than the official
sources). I think, that the group who initiated the mapping should
'clean up after themselves' (and I just wanted to affirm that we will do
that). I just wanted to express gratitude to mappers how do help out.

Once I am more into it, I am happy to help out validating other's work.

I did not mean to cheery pick when I quoted the validation website (I
very much appreciate the wiki page). I just wanted to make a point about
timing.

Regarding Canada, as a geographer I am fully aware of the fact Canada
having relatively less population, however, it has still almost half of
the population of Germany and the urban areas, which most of OSM mappers
are concerned with, might be relatively (to population) similar in size
(that's just a guess). I would have to run a query now to find out if
the relative number of active mappers is higher in one country than the
other, but that's not my point. The relative numbers do not matter, as
actual people do the mapping. And there, I hope we agree, the Canadian
OSM community could do with more active mappers.

Tim


Am 23.11.2017 um 07:54 schrieb john whelan:
The issue is the quality of the mapping, nothing else. I attended one of
these geoweek events and we used JOSM with the building_tool plugin. 
The mapping of buildings was accurate even though 75% of the mappers had
never used JOSM before.  There was no formal validation done but I
verified each mappers work as they did it.  I got the impression that
the mappers enjoyed the exercise and I think for me that was the most
important thing.  Mapping should be fun.

There was no mention of the work would be validated nor did we record
the mappers userids to ensure which mappers had mapped.  Other mappers
had marked tiles done on the grid.

I was under the impression that Stats Canada was involved but was later
assured by them that this was not the case.

The problem of lots of new mappers producing low quality work really
reared its head during the Nepal crisis.  I do mainly validation on HOT
projects in Africa and I ended up pulling in chunks of Africa and just
trying to clean up the map.  Currently I'm looking at one mapper who has
added more than a thousand ways with one tag I think it says
source=PGS.  Data quality is a major issue in OpenStreetMap.  Recently
someone gave up when looking for area=yes or buildings drawn in iD but
left untagged for the most part.  I think in Europe it was 100,000 or
more worldwide it was far higher and that's when the person looking at
it gave up.

There are many examples in Africa of groups of buildings being mapped as
one building and labelled building=house.  That's what we are trying to
avoid.  It is possible to correctly map a building in iD I've seen it
done but it takes time.  It is far easier to sort of roughly get it
right and roughly means not accurately.  I think the thing we need to
avoid is a feeling the mapper needs to get a tile done. That's when they
start to rush things.

Building validation?  I can think of no validator who enjoys having to
take two or three times longer to correct someones's work than it would
take them to map it in JOSM with the building_tool in the first place. 
I'm unable to even think of a case where a project has been validated
and the buildings corrected.  When I validate I'm trying to correct the
mapper's work and give them feedback so they will map more accurately in
future.  There is no point in doing this to someone who will map once. 
It's a waste of my time.

The wiki page you pointed to, I wrote much of it. the most important
part which you skipped is feedback from a user. 


  Why do we validate?

“OpenStreetMap is often the only source of maps, but the data quality is
very uneven.  I wish they’d put their more experienced mappers onto
validation.”  This is a quote from an individual who used OpenStreetMap
data (HOT) in the field.

Note the comment the data quality is very uneven and that's what we are
trying to address.  Your particular maperthon may have produced good
work, my lot certainly did but many mappers using the tag did not and
that is the issue.

By the way we do have fewer mappers per square kilometre than Germany
does and we have used CANVEC data to get a basic road network in.  In
Ottawa we've used Open Data to bring in the bus stops. The basic
Canadian map isn't bad but if we had as many mappers per square
kilometre as Germany does then no doubt 

Re: [Talk-ca] Mapping buildings in Canada by 2020

[john whelan]

The issue is the quality of the mapping, nothing else. I attended one of
these geoweek events and we used JOSM with the building_tool plugin.  The
mapping of buildings was accurate even though 75% of the mappers had never
used JOSM before.  There was no formal validation done but I verified each
mappers work as they did it.  I got the impression that the mappers enjoyed
the exercise and I think for me that was the most important thing.  Mapping
should be fun.

There was no mention of the work would be validated nor did we record the
mappers userids to ensure which mappers had mapped.  Other mappers had
marked tiles done on the grid.

I was under the impression that Stats Canada was involved but was later
assured by them that this was not the case.

The problem of lots of new mappers producing low quality work really reared
its head during the Nepal crisis.  I do mainly validation on HOT projects
in Africa and I ended up pulling in chunks of Africa and just trying to
clean up the map.  Currently I'm looking at one mapper who has added more
than a thousand ways with one tag I think it says source=PGS.  Data quality
is a major issue in OpenStreetMap.  Recently someone gave up when looking
for area=yes or buildings drawn in iD but left untagged for the most part.
I think in Europe it was 100,000 or more worldwide it was far higher and
that's when the person looking at it gave up.

There are many examples in Africa of groups of buildings being mapped as
one building and labelled building=house.  That's what we are trying to
avoid.  It is possible to correctly map a building in iD I've seen it done
but it takes time.  It is far easier to sort of roughly get it right and
roughly means not accurately.  I think the thing we need to avoid is a
feeling the mapper needs to get a tile done. That's when they start to rush
things.

Building validation?  I can think of no validator who enjoys having to take
two or three times longer to correct someones's work than it would take
them to map it in JOSM with the building_tool in the first place.  I'm
unable to even think of a case where a project has been validated and the
buildings corrected.  When I validate I'm trying to correct the mapper's
work and give them feedback so they will map more accurately in future.
There is no point in doing this to someone who will map once.  It's a waste
of my time.

The wiki page you pointed to, I wrote much of it. the most important part
which you skipped is feedback from a user.

Why do we validate?

“OpenStreetMap is often the only source of maps, but the data quality is
very uneven.  I wish they’d put their more experienced mappers onto
validation.”  This is a quote from an individual who used OpenStreetMap
data (HOT) in the field.
Note the comment the data quality is very uneven and that's what we are
trying to address.  Your particular maperthon may have produced good work,
my lot certainly did but many mappers using the tag did not and that is the
issue.

By the way we do have fewer mappers per square kilometre than Germany does
and we have used CANVEC data to get a basic road network in.  In Ottawa
we've used Open Data to bring in the bus stops. The basic Canadian map
isn't bad but if we had as many mappers per square kilometre as Germany
does then no doubt it would be better.  Our population density is also
lower by the way if you hadn't noticed.

Cheerio John

On 22 November 2017 at 21:28, Tim Elrick  wrote:

> Hello all,
>
> As you know Open Mapping Group McGill (OMG McGill) organized one of the
> mapathons last week for the town of Williams Lake, BC. For the turnout
> please turn to Julia's website published earlier today on the list.
>
> As a mentor of the group I might be the 'director' of this event
> according to the proposed policy by the OSMF board. In this role, I want
> to assure you that we tried to do our best to teach new mappers how to
> do their job properly, as Charles stated on this list yesterday. And
> judging from a preliminary analysis of the data I conducted with the
> overpass api, the participants did a pretty good job.
>
> Of course, the data needs validation, which we will conduct in the next
> couple of days. However, I do not see the rush proposed on this list
> earlier. Ideally, validation would happen right after the mapping event
> (as set out in this manual for HOT tasks [1]). In the real world, we all
> have our jobs, families and other voluntary engagements, that sometimes
> do not allow to act accordingly. I further think it is not even
> necessary for tasks that are not related to immediate disaster response
> or include ways tagged with a highway tag (in the later case it might
> confuse navigation apps if not validated right away). In many cases,
> validation, or better, correction of data entered by individual mappers
> (not part of group events) was (and still is) done many days or even
> months after the data was entered, depending on whether an experienced
> mapper has 

[Talk-ca] Mapping buildings in Canada by 2020

[Tim Elrick]

Hello all,

As you know Open Mapping Group McGill (OMG McGill) organized one of the
mapathons last week for the town of Williams Lake, BC. For the turnout
please turn to Julia's website published earlier today on the list.

As a mentor of the group I might be the 'director' of this event
according to the proposed policy by the OSMF board. In this role, I want
to assure you that we tried to do our best to teach new mappers how to
do their job properly, as Charles stated on this list yesterday. And
judging from a preliminary analysis of the data I conducted with the
overpass api, the participants did a pretty good job.

Of course, the data needs validation, which we will conduct in the next
couple of days. However, I do not see the rush proposed on this list
earlier. Ideally, validation would happen right after the mapping event
(as set out in this manual for HOT tasks [1]). In the real world, we all
have our jobs, families and other voluntary engagements, that sometimes
do not allow to act accordingly. I further think it is not even
necessary for tasks that are not related to immediate disaster response
or include ways tagged with a highway tag (in the later case it might
confuse navigation apps if not validated right away). In many cases,
validation, or better, correction of data entered by individual mappers
(not part of group events) was (and still is) done many days or even
months after the data was entered, depending on whether an experienced
mapper has an eye on a certain region or not. With regards to buildings
in areas where there existed no respective data before, I do not see any
need for rushing.

The important thing is that the organiser of a group event makes sure
that the data entered by participants of the event *is* validated to
ensure data quality. And we will. To this end, I appreciate that
long-term members already offered to help us there (thank you, Charles!).

I still consider mapathons a legitimate way to draw attention to OSM, to
advocate for open data, and to show the potential of OSM data and the
lack thereof in many parts of the world, including Canada. From the
experience of our first mapathon I got the impression that we instigated
a vast interest in open mapping (which, I think, is a valid goal on its
own right) and I expect quite a couple of returning participants to our
next events, in which we will train them further on the complexities to
produce good OSM data. By continuing, we might be able to motivate one
or two persons to turn into long-term mappers; this is, by the way,
totally in line with the long-tail phenomenon researchers found in all
crowd-sourcing projects.
All those reasons I mentionend, are, I think, worth it continuing doing
what we did. I would appreciate, if the attitude towards group mapping
events were less hostile on this list and on OSM as such (I am aware of
less fortunate attempts conducting group mapping events recently; but
try not blame them, but give them a hand to do it better next time - and
I know you did, but some of them apparently did not understand how
communication works in OSM). Try to give them the benefit of the doubt:
most mappers, even in group event, do this voluntarily and because they
want to enjoy extend this great geodatabase!

IMHO, OSM cannot do without those events, because we do not want to
leave the future of OSM only to businesses and their paid mappers (and
we have seen that in some countries, including Canada, there might not
be enough people who find their way to OSM without those events).

Tim


[1]
http://wiki.openstreetmap.org/wiki/OSM_Tasking_Manager/Validating_data#When_do_we_validate.3F


___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Mapping buildings in Canada by 2020

[Julia C]

Hello everyone!

I am Julia Conzon, username Noznoc. I am currently *interning* at Mapbox
until the end of November.

My involvement in BC2020 came from my past work at Statistics Canada on the
Crowdsourcing project and my interest in VGI. *I am not a representative
for Mapbox (Mapbox is only supporting me)*.

Organizing these mapathons have been a great experience because I have
witnessed academics and students across Canada interested in being educated
on OSM and contributing to the OSM project through BC2020. Several
universities intend on continuing mapathons.

I have just finished developing a visualization
 that
showcases the results from the mapathon (data collected between Nov.
10-20). To learn more about the visualization, I recommend reading the
README.md file in the GitHub repo
.

>From what Pierre's email states, I guess the validation work flow panned
out from OSMGeoWeek does not follow the normal script that has been done in
the past, *but that doesn't mean it doesn't exist*. The universities are
aware of validation, and have the intentions to work on it; however, since
involvement is volunteered-based, people have other priorities, so
validation will not occur immediately. My visualization is developed in a
way to support validation efforts, and I have started documenting
validation methods here

 (*feel free to add to the wiki*).

>From my time at Mapbox I have learned methods for validating data, and
developing front-end apps to easily see and fix errors. I would love to
collaborate with any other individuals interested in developing these tools
for BC2020. This would be on my own time, this is not a part of my job, it
is part of my hobby.

Regards,
Julia


On Mon, Nov 20, 2017 at 5:19 PM, john whelan  wrote:

> >BC2020 *is not a StatCan project*.
>
> That comment came from Bjenk's old boss.
>
> I note that Pierre has identified some data quality issues from maperthons
> that appear to be associated with this.
>
> Is anyone organising this or is it just a dream in the air?
>
> In Ottawa we got some high quality mapping out of the initial Stats Canada
> project which is good.
>
> The issue on the low quality mapping by mappers who will map once then
> disappear is what if anything should be done about the less than ideal
> mapping left behind?
>
> Traditionally its been suggested that is it best corrected but it takes
> longer to correct than to delete and remap.
>
> It doesn't seem to be easy to handle at a local level.  There is too much
> just dropped on one spot at once.
>
> One mapper made a comment to me on this type of mapping in Africa just
> delete this junk.  I have some sympathy with this point of view.
>
> Thoughts if any
>
> Thanks John
>
>
>
> ___
> Talk-ca mailing list
> Talk-ca@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ca
>
>
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

Re: [Talk-ca] Mapping buildings in Canada by 2020

[Pierre Béland]

Bonjour Charles,
Je suis en contact avec le responsable de McGill. Effectivement, dans leurs cas 
ils prévoient réviser le travail avec les étudiants.
Par contre dans le cadre de Mapathons, il est important de prévoir en général 
de bien sopporter les nouveaux contributeurs. Il me semble important d'avoir 
des outils de suivi des contributions et corriger très rapidement les 
problèmes. Des requêtes Overpass peuvent être utilisées pour suivre l'ensemble 
du groupe ou des contributeurs individuels. La possibilité de projeter sur 
écran peut aussi grandement faciliter l'interaction avec les participants.


 
Pierre 
 

Le lundi 20 novembre 2017 21:17:42 HNE, Charles Basenga Kiyanda 
 a écrit :  
 
 Bonjour Pierre,

Juste quelques précisions car j'ai été impliqué dans une de ces activités. Dans 
le cas de Williams Lake (projet 91), il a été utilisé pour un mapathon d'un 
nouveau groupe (Open Mapping Group McGill). C'est un nouveau groupe, mais qui 
me semble très bien organisé. Il ya eu trois présentations qui, en tout,, ont 
duré environ 1h15,, avant que les participants se mettent à contribuer. 

Le projet sur le gestionnaire de tâches à été créé par Julia Conzon (sp?) de 
Mapbox qui a donné une des conférences par vidéo. J'assume que c'est 
l'utilisateur noznoc.

Pour ce qui est de cet événement en particulier, c'est vrai que la partie 
validation tarde, mais je crois bien qu'elle se fera. Leur groupe semble très 
bien organisé. Il émane surtout du département de géographie avec un bon 
support de membres du personnel. 

Je n'extrapolerai pas aux autres activités et je commentterai pas sur les 
techniques de Mapbox, mais pour ce qui est de l'activité à McGill liée à 
williams Lake, c'était une activité bien organisée.

Maintenant, il reste à structurer la validation des données. C'est sûr que sans 
l'urgence d'un projet HOT, c'est plus difficile de motiver les gens à continuer 
leur implication.

Charles

On November 20, 2017 8:56:18 PM EST, "Pierre Béland"  wrote:
John
Je t'ai informé précédemment que je constatais pour Calgary des doublons 
d'adresse, celle-ci était ajoutée à la fois sur l'immeuble principal et le 
garage. Et vérifiant rapidement, j'ai vu des cas similaires à Ottawa. Je n'ai 
cependant pas analysé plus en détail la qualité des éditions pour le projet 
bc2020. Cependant, il ne faut pas oublier qu'il est important d'assurer un bon 
suivi des nouveaux. Souvent ils se concentrent sur les immeubles mais 
produisent des résultats très imprécis. Il est important de les accompagner et 
corriger rapidement. Sinon, les contributeurs plus expérimentés éviterons 
ensuite ces zones avec autant de problèmes à corriger.
Rappellons que depuis le début, ce sont des représentants de MapBox qui ont 
présenté ce projet et créé une certaine confusion sur le rôle de la communauté 
OSM du Canada. Nous avons été très clairs que ce projet n'est pas une priorité 
pour notre communatué. Ils sont revenus à la charge avec la semaine Geoweek, 
mettant encore de l'avant le projet bc2020 et invitant les universités à y 
participer. Je suis en désaccord avec ce mode de fonctionnement des partenaires 
corporatifs qui prennent des décisions au nom de notre communauté et n'assurent 
pas ensuite les responsabilités reliées à de tels projets.
J'ai analysé le travail effectué par les contributeurs OSM dans le cadre de la 
semaine GeoWeek (données du 12 au 20 novembre). Pour le projet bc2020, sept 
projets ont été créés sur le serveur tasks.osmcanada.ca par le contributeur 
Noznoc. Je ne saurais dire s'il est lié à MapBox. Au Canada, 167 contributeurs 
ont édité 82,570 objets (environ 16,500 immeubles, 494 immeubles par personne). 
Parmi eux, 96 nouveaux contributeurs depuis le début de la semaine dernière ont 
édité 38,211 objets (environ 7,640 immeubles, 80 immeubles par personne). Comme 
John l'a souvent souligné, il est important de repérer rapidement les 
problèmes. Sinon les contributeurs expérimentés éviterons ensuite d'éditer ces 
zones, puisqu'il est plus long de corriger ces données tout en conservant 
l'historique que de partir d'une feuille vierge et y ajouter de nouveaux 
immeubles à l'aide du greffon immeubles de JOSM.

A part d'inviter les universités à participer, MapBox n'a assuré aucun suivi au 
cours de la semaine et la validation du travail des nouveaux participants a été 
quasiment nulle si on se fie aux statistiques du serveur de tâches (quelques 
carreaux en vert).
Aussi pour le suivi de tels projets, il est important de bien documenter les 
projets sur le gestionnaire de tâches. Dans les instructions de différents 
projets listés ci-dessous, aucune référence n'est ajoutée. Idéalement, ont 
devrait y retrouver une référence du style #osmcanada-xx (xx=projet). J'ai pu 
néanmoins déterminer dans la base de données des Changesets les statistiques de 
contributions au projet #bc2020 présentées ci-haut.

Projets sur tasks.osmcanada.ca83 Calgary
85 Durham 

Re: [Talk-ca] Mapping buildings in Canada by 2020

[Charles Basenga Kiyanda]

Bonjour Pierre,

Juste quelques précisions car j'ai été impliqué dans une de ces activités. Dans 
le cas de Williams Lake (projet 91), il a été utilisé pour un mapathon d'un 
nouveau groupe (Open Mapping Group McGill). C'est un nouveau groupe, mais qui 
me semble très bien organisé. Il ya eu trois présentations qui, en tout,, ont  
duré  environ 1h15,, avant que les participants se mettent à contribuer. 

Le projet sur le gestionnaire de tâches à été créé par Julia Conzon (sp?) de 
Mapbox qui a donné une des conférences par vidéo. J'assume que c'est 
l'utilisateur noznoc.

Pour ce qui est de cet événement en particulier, c'est vrai que la partie 
validation tarde, mais je crois bien qu'elle se fera. Leur groupe semble très 
bien organisé. Il émane surtout du département de géographie avec un bon 
support de membres du personnel. 

Je n'extrapolerai pas aux autres activités et je commentterai pas sur les 
techniques de Mapbox, mais pour ce qui est de l'activité à McGill liée à 
williams Lake, c'était une activité bien organisée.

Maintenant, il reste à structurer la validation des données. C'est sûr que sans 
l'urgence d'un projet HOT, c'est plus difficile de motiver les gens à continuer 
leur implication.

Charles

On November 20, 2017 8:56:18 PM EST, "Pierre Béland"  wrote:
>John
>Je t'ai informé précédemment que je constatais pour Calgary des
>doublons d'adresse, celle-ci était ajoutée à la fois sur l'immeuble
>principal et le garage. Et vérifiant rapidement, j'ai vu des cas
>similaires à Ottawa. Je n'ai cependant pas analysé plus en détail la
>qualité des éditions pour le projet bc2020. Cependant, il ne faut pas
>oublier qu'il est important d'assurer un bon suivi des nouveaux.
>Souvent ils se concentrent sur les immeubles mais produisent des
>résultats très imprécis. Il est important de les accompagner et
>corriger rapidement. Sinon, les contributeurs plus expérimentés
>éviterons ensuite ces zones avec autant de problèmes à corriger.
>Rappellons que depuis le début, ce sont des représentants de MapBox qui
>ont présenté ce projet et créé une certaine confusion sur le rôle de la
>communauté OSM du Canada. Nous avons été très clairs que ce projet
>n'est pas une priorité pour notre communatué. Ils sont revenus à la
>charge avec la semaine Geoweek, mettant encore de l'avant le projet
>bc2020 et invitant les universités à y participer. Je suis en désaccord
>avec ce mode de fonctionnement des partenaires corporatifs qui prennent
>des décisions au nom de notre communauté et n'assurent pas ensuite les
>responsabilités reliées à de tels projets.
>J'ai analysé le travail effectué par les contributeurs OSM dans le
>cadre de la semaine GeoWeek (données du 12 au 20 novembre). Pour le
>projet bc2020, sept projets ont été créés sur le serveur
>tasks.osmcanada.ca par le contributeur Noznoc. Je ne saurais dire s'il
>est lié à MapBox. Au Canada, 167 contributeurs ont édité 82,570 objets
>(environ 16,500 immeubles, 494 immeubles par personne). Parmi eux, 96
>nouveaux contributeurs depuis le début de la semaine dernière ont édité
>38,211 objets (environ 7,640 immeubles, 80 immeubles par personne).
>Comme John l'a souvent souligné, il est important de repérer rapidement
>les problèmes. Sinon les contributeurs expérimentés éviterons ensuite
>d'éditer ces zones, puisqu'il est plus long de corriger ces données
>tout en conservant l'historique que de partir d'une feuille vierge et y
>ajouter de nouveaux immeubles à l'aide du greffon immeubles de JOSM.
>
>A part d'inviter les universités à participer, MapBox n'a assuré aucun
>suivi au cours de la semaine et la validation du travail des nouveaux
>participants a été quasiment nulle si on se fie aux statistiques du
>serveur de tâches (quelques carreaux en vert).
>Aussi pour le suivi de tels projets, il est important de bien
>documenter les projets sur le gestionnaire de tâches. Dans les
>instructions de différents projets listés ci-dessous, aucune référence
>n'est ajoutée. Idéalement, ont devrait y retrouver une référence du
>style #osmcanada-xx (xx=projet). J'ai pu néanmoins déterminer dans la
>base de données des Changesets les statistiques de contributions au
>projet #bc2020 présentées ci-haut.
>
>Projets sur tasks.osmcanada.ca83 Calgary
>85 Durham Region
>86 Stratford
>87 Trail
>88 Massett
>91 Williams Lake
>92 North Battleford
> 
>Pierre 
> 
>
>Le lundi 20 novembre 2017 17:20:20 HNE, john whelan
> a écrit :  
> 
> ___
>Talk-ca mailing list
>Talk-ca@openstreetmap.org
>https://lists.openstreetmap.org/listinfo/talk-ca
>  >BC2020is not a StatCan project.
>That comment came from Bjenk's old boss.
>I note that Pierre has identified some data quality issues from
>maperthons that appear to be associated with this.
>
>Is anyone organising this or is it just a dream in the air?
>
>In Ottawa we got some high quality mapping out of the initial Stats
>Canada project which is good.
>
>The issue on the low 

Re: [Talk-ca] Mapping buildings in Canada by 2020

[Pierre Béland]

John
Je t'ai informé précédemment que je constatais pour Calgary des doublons 
d'adresse, celle-ci était ajoutée à la fois sur l'immeuble principal et le 
garage. Et vérifiant rapidement, j'ai vu des cas similaires à Ottawa. Je n'ai 
cependant pas analysé plus en détail la qualité des éditions pour le projet 
bc2020. Cependant, il ne faut pas oublier qu'il est important d'assurer un bon 
suivi des nouveaux. Souvent ils se concentrent sur les immeubles mais 
produisent des résultats très imprécis. Il est important de les accompagner et 
corriger rapidement. Sinon, les contributeurs plus expérimentés éviterons 
ensuite ces zones avec autant de problèmes à corriger.
Rappellons que depuis le début, ce sont des représentants de MapBox qui ont 
présenté ce projet et créé une certaine confusion sur le rôle de la communauté 
OSM du Canada. Nous avons été très clairs que ce projet n'est pas une priorité 
pour notre communatué. Ils sont revenus à la charge avec la semaine Geoweek, 
mettant encore de l'avant le projet bc2020 et invitant les universités à y 
participer. Je suis en désaccord avec ce mode de fonctionnement des partenaires 
corporatifs qui prennent des décisions au nom de notre communauté et n'assurent 
pas ensuite les responsabilités reliées à de tels projets.
J'ai analysé le travail effectué par les contributeurs OSM dans le cadre de la 
semaine GeoWeek (données du 12 au 20 novembre). Pour le projet bc2020, sept 
projets ont été créés sur le serveur tasks.osmcanada.ca par le contributeur 
Noznoc. Je ne saurais dire s'il est lié à MapBox. Au Canada, 167 contributeurs 
ont édité 82,570 objets (environ 16,500 immeubles, 494 immeubles par personne). 
Parmi eux, 96 nouveaux contributeurs depuis le début de la semaine dernière ont 
édité 38,211 objets (environ 7,640 immeubles, 80 immeubles par personne). Comme 
John l'a souvent souligné, il est important de repérer rapidement les 
problèmes. Sinon les contributeurs expérimentés éviterons ensuite d'éditer ces 
zones, puisqu'il est plus long de corriger ces données tout en conservant 
l'historique que de partir d'une feuille vierge et y ajouter de nouveaux 
immeubles à l'aide du greffon immeubles de JOSM.

A part d'inviter les universités à participer, MapBox n'a assuré aucun suivi au 
cours de la semaine et la validation du travail des nouveaux participants a été 
quasiment nulle si on se fie aux statistiques du serveur de tâches (quelques 
carreaux en vert).
Aussi pour le suivi de tels projets, il est important de bien documenter les 
projets sur le gestionnaire de tâches. Dans les instructions de différents 
projets listés ci-dessous, aucune référence n'est ajoutée. Idéalement, ont 
devrait y retrouver une référence du style #osmcanada-xx (xx=projet). J'ai pu 
néanmoins déterminer dans la base de données des Changesets les statistiques de 
contributions au projet #bc2020 présentées ci-haut.

Projets sur tasks.osmcanada.ca83 Calgary
85 Durham Region
86 Stratford
87 Trail
88 Massett
91 Williams Lake
92 North Battleford
 
Pierre 
 

Le lundi 20 novembre 2017 17:20:20 HNE, john whelan  
a écrit :  
 
 ___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca
  >BC2020is not a StatCan project.
That comment came from Bjenk's old boss.
I note that Pierre has identified some data quality issues from maperthons that 
appear to be associated with this.

Is anyone organising this or is it just a dream in the air?

In Ottawa we got some high quality mapping out of the initial Stats Canada 
project which is good.

The issue on the low quality mapping by mappers who will map once then 
disappear is what if anything should be done about the less than ideal mapping 
left behind?
Traditionally its been suggested that is it best corrected but it takes longer 
to correct than to delete and remap.

It doesn't seem to be easy to handle at a local level.  There is too much just 
dropped on one spot at once.
One mapper made a comment to me on this type of mapping in Africa just delete 
this junk.  I have some sympathy with this point of view.

Thoughts if any
Thanks John


___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca

[Talk-ca] Mapping buildings in Canada by 2020

[john whelan]

>BC2020 *is not a StatCan project*.

That comment came from Bjenk's old boss.

I note that Pierre has identified some data quality issues from maperthons
that appear to be associated with this.

Is anyone organising this or is it just a dream in the air?

In Ottawa we got some high quality mapping out of the initial Stats Canada
project which is good.

The issue on the low quality mapping by mappers who will map once then
disappear is what if anything should be done about the less than ideal
mapping left behind?

Traditionally its been suggested that is it best corrected but it takes
longer to correct than to delete and remap.

It doesn't seem to be easy to handle at a local level.  There is too much
just dropped on one spot at once.

One mapper made a comment to me on this type of mapping in Africa just
delete this junk.  I have some sympathy with this point of view.

Thoughts if any

Thanks John
___
Talk-ca mailing list
Talk-ca@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk-ca