[Talk-ca] Talk-ca Digest, Vol 96, Issue 1

Steve Singer steve at ssinger.info
Wed Feb 3 00:54:26 UTC 2016


On Tue, 2 Feb 2016, Mojgan Jadidi wrote:

Mojgan,

We did say at MappyHour last month that OSM was filled with well intentioned 
imports gone bad. Don't feel bad you seem to be well intentioned and many 
others before you have made mistakes in the import process.

I don't think the advice you got at the meeting was to go rush ahead and 
start importing. I think the advice you received was more along the lines of 
"imports are really hard, getting community consensus is even harder", 
"spend some time understanding how the OSM community works first", "use 
dedicated import accounts attached to individuals not an organization"

Some specific feedback I have on the import

1) I think it is critical that you do the data validation and checking 
*before* uploading to OSM.  This is what I thought you had described but I 
must have misunderstood.  The reason why it is better to do the cleanup 
before uploading the data to OSM is because this way no bad data is sitting 
in OSM until someone gets around to cleaning it up.  The community also 
likes this better because we don't have to worry about if you will ever get 
around to cleaning it up

The flow I would recommend is along the lines
1) Person takes a small area of generated address points
2) They download the current OSM data and check for problems
3) They fix any issues and upload the small area they just checked

The changesets on Monday appeared to be uploading everything first.


2) You should publish your generated files for review(or at least a sample). 
In some areas I spot checked last night I saw nodes with no tags and not 
connected to an interpolation way.  This might have been an artifact of a 
in-progress revert, I'm not sure.

I also saw some places where the existing OSM roads and address 
interpolations were a bit offset from the stuff you uploaded and this 
resulted in duplicate address ways (the existing geometry in OSM might not 
be from a rough survey). Manual cleanup would be required.


3) I am not sure if we need the :source tag on each node or if putting it on 
the changeset would be enough.  Earlier imports tended to put metadata on 
each node but I think we have been trying to move away from this.


Steve


> Hi all,Please accept my apology for this misunderstanding, I thought our presence for last OSM meeting and
> flowing emails in Talk-ca and with John were part of communication and discussion, we launched our wikipage
> two weeks ago, and we did not receive any feedback, so we thought to start import via JOSM. 
> 
> as we expalined in our wikipage, we created the algorithm to detect the missing information, and then we
> check the quality of this information on JOSM on the top of OpenStreetMap, Bing Areal imagery, GeoBase Road
> network and some local municpal open data. the data is created initially through StatCan, however, we noticed
> the low quality of StatCan road segment geometry, so we deal with this issue by using complement dataset such
> as OpenStreetMap, Bing Areal imagery, GeoBase Road network and some local municpal open data. all created
> nodes and ways are carefully inspected visually using above dataset for more that 6 weeks. 
> 
> Our final verification will be on the OSM server (on-line) to avoid or detect any issues. Our aim is having
> high quality address information in OSM for sake of community. we were very prudent from the first step to
> have a high quality source of information. 
> 
> I hope that the community will accept our contribution and enjoy to use the data.
> 
> Best regards,
> 
> Mojgan  
>   
> 
> Mojgan (Amaneh) Jadidi, Ph.D.
> Postdoctoral Research Fellow
> GeoICT Lab
> York University
> Toronto
> 
> ca.linkedin.com/pub/mojgan-amaneh-jadidi/10/825/969/
> 
> On 2 February 2016 at 07:00, <talk-ca-request at openstreetmap.org> wrote:
>       Send Talk-ca mailing list submissions to
>               talk-ca at openstreetmap.org
>
>       To subscribe or unsubscribe via the World Wide Web, visit
>               https://lists.openstreetmap.org/listinfo/talk-ca
>       or, via email, send a message with subject or body 'help' to
>               talk-ca-request at openstreetmap.org
>
>       You can reach the person managing the list at
>               talk-ca-owner at openstreetmap.org
>
>       When replying, please edit your Subject line so it is more specific
>       than "Re: Contents of Talk-ca digest..."
> 
>
>       Today's Topics:
>
>          1. Triplinx import (Stewart Russell)
>          2. Re: Triplinx import (john whelan)
>          3. Re: Triplinx import (Stewart Russell)
> 
>
>       ----------------------------------------------------------------------
>
>       Message: 1
>       Date: Mon, 1 Feb 2016 18:36:01 -0500
>       From: Stewart Russell <scruss at gmail.com>
>       To: talk-ca <talk-ca at openstreetmap.org>
>       Subject: [Talk-ca] Triplinx import
>       Message-ID:
>               <CAAsTreDSk=ZPf5zniBH144N46pMJgcvX8q-vdYHRMwjJ188ohw at mail.gmail.com>
>       Content-Type: text/plain; charset="utf-8"
>
>       It seems that this import has started with no discussion. Here's the wiki
>       page:
>       https://wiki.openstreetmap.org/wiki/Triplinx_Metrolinx_Import_Plan
>
>       Stewart
>       -------------- next part --------------
>       An HTML attachment was scrubbed...
>       URL:
>       <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20160201/2432eb78/attachment-0001.html>
>
>       ------------------------------
>
>       Message: 2
>       Date: Mon, 1 Feb 2016 19:14:11 -0500
>       From: john whelan <jwhelan0112 at gmail.com>
>       To: Stewart Russell <scruss at gmail.com>, Mojgan Jadidi
>               <mojgan.jadidi at gmail.com>
>       Cc: talk-ca <talk-ca at openstreetmap.org>
>       Subject: Re: [Talk-ca] Triplinx import
>       Message-ID:
>               <CAJ-Ex1GJSvCrz4+j4cq=rDZo_WD8dB6q+YvTOMMsgQEyuem=bw at mail.gmail.com>
>       Content-Type: text/plain; charset="utf-8"
>
>       I had a skype call with Mojgan where we discussed some of the issues and
>       there was an email sent to talk-ca on Jan 12th 2016.
>
>       To me a couple of issues first are that the Stats Canada data was designed
>       for Stats Canada use, probably for the labour force survey and as such
>       provided the interviewer can see the address from where it is marked on the
>       map this is sufficent accuracy.  My concern would be about the accuracy of
>       the data.
>
>       Another issue is do we end up with duplicate addresses?  My concern would
>       be do we end up with the same address in two different locations?
>
>       I think a better solution would be to take a copy of the OSM database,
>       strip out the existing address data and drop in the stats data and use the
>       copy for their purposes.
>
>       Cheerio John
>
>       On 1 February 2016 at 18:36, Stewart Russell <scruss at gmail.com> wrote:
>
>       > It seems that this import has started with no discussion. Here's the wiki
>       > page:
>       > https://wiki.openstreetmap.org/wiki/Triplinx_Metrolinx_Import_Plan
>       >
>       > Stewart
>       >
>       > _______________________________________________
>       > Talk-ca mailing list
>       > Talk-ca at openstreetmap.org
>       > https://lists.openstreetmap.org/listinfo/talk-ca
>       >
>       >
>       -------------- next part --------------
>       An HTML attachment was scrubbed...
>       URL:
>       <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20160201/380bb9cc/attachment-0001.html>
>
>       ------------------------------
>
>       Message: 3
>       Date: Mon, 1 Feb 2016 19:38:19 -0500
>       From: Stewart Russell <scruss at gmail.com>
>       To: talk-ca <talk-ca at openstreetmap.org>
>       Subject: Re: [Talk-ca] Triplinx import
>       Message-ID:
>               <CAAsTreC+jpJdAOeEG7z2GX95T4tjm=GdT=ms4t7iDXf4tx7YEQ at mail.gmail.com>
>       Content-Type: text/plain; charset="utf-8"
>
>       Import user is:
>       https://www.openstreetmap.org/user/Triplinx%20Canada
>
>       Several 50K node imports today.
>
>       Stewart
>       On Feb 1, 2016 7:14 PM, "john whelan" <jwhelan0112 at gmail.com> wrote:
>
>       > I had a skype call with Mojgan where we discussed some of the issues and
>       > there was an email sent to talk-ca on Jan 12th 2016.
>       >
>       > To me a couple of issues first are that the Stats Canada data was designed
>       > for Stats Canada use, probably for the labour force survey and as such
>       > provided the interviewer can see the address from where it is marked on the
>       > map this is sufficent accuracy.  My concern would be about the accuracy of
>       > the data.
>       >
>       > Another issue is do we end up with duplicate addresses?  My concern would
>       > be do we end up with the same address in two different locations?
>       >
>       > I think a better solution would be to take a copy of the OSM database,
>       > strip out the existing address data and drop in the stats data and use the
>       > copy for their purposes.
>       >
>       > Cheerio John
>       >
>       > On 1 February 2016 at 18:36, Stewart Russell <scruss at gmail.com> wrote:
>       >
>       >> It seems that this import has started with no discussion. Here's the wiki
>       >> page:
>       >> https://wiki.openstreetmap.org/wiki/Triplinx_Metrolinx_Import_Plan
>       >>
>       >> Stewart
>       >>
>       >> _______________________________________________
>       >> Talk-ca mailing list
>       >> Talk-ca at openstreetmap.org
>       >> https://lists.openstreetmap.org/listinfo/talk-ca
>       >>
>       >>
>       >
>       -------------- next part --------------
>       An HTML attachment was scrubbed...
>       URL:
>       <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20160201/d27749c7/attachment-0001.html>
>
>       ------------------------------
>
>       Subject: Digest Footer
>
>       _______________________________________________
>       Talk-ca mailing list
>       Talk-ca at openstreetmap.org
>       https://lists.openstreetmap.org/listinfo/talk-ca
> 
>
>       ------------------------------
>
>       End of Talk-ca Digest, Vol 96, Issue 1
>       **************************************
> 
> 
> 
>


More information about the Talk-ca mailing list