[talk-au] Investigating a NSW Schools Import

Andrew Harvey andrew.harvey4 at gmail.com
Sun Apr 29 13:49:05 UTC 2018


I've done this import, where existing data conflicts I've mostly left in
place, sometimes when it's obviously wrong I've fixed it.

There were a lot of intstances where some data existed by in a similar tag,
eg. website, contact:website and url. As much as possible I've tried to
leave in place the existing key.

A few were omitted as I lack the local knowledge I feel necessary to add
them to OSM: Aurora College, Saturday School of Community Languages,
Central Sydney Intensive English High School, plus some others where I've
left notes.

All commits happened in https://www.openstreetmap.org/user/aharvey_imports


On 28 April 2018 at 11:20, Andrew Harvey <andrew.harvey4 at gmail.com> wrote:

> I've received some feedback from the imports list and made some changes to
> the plan.
>
> 1. The ref tags won't be imported as they can in some instances discourage
> local edits. They make the object feel less like a native OSM object and
> editors may feel it's a special imported piece of data that they can't
> touch. I certainly don't want this imported data to feel like that, and
> given these IDs aren't that important (feel free to identify any use cases
> where they are helpful) I'll skip them initially.
>
> 2. access=private tag won't be imported, this is best surveyed on an
> individual basis.
>
> 3. capacity changed to school:enrolment to better reflect what this number
> actually is. Enrolment could be lower or greater than the capacity.
>
> 4. I'll put tho source tags on the changeset, moving existing source tags
> to source:geometry where it is an Imagery source, or if source is survey,
> leaving it in place.
>
> 5. I'll leave in place existing "Foo Primary School" names as what's on
> the ground should prevail. "Foo Public School" where "Foo Primary School"
> exists will be placed in the alt_name. These can always be changed on an
> individual basis with local knowledge of the school.
>
> I'll give it a few more days for feedback, if I don't get anything back
> I'll start the import as planned.
>
> On 23 April 2018 at 21:19, Andrew Harvey <andrew.harvey4 at gmail.com> wrote:
>
>> I'm investigating the possibility of importing NSW Public Schools data
>> <https://data.cese.nsw.gov.au/data/dataset/nsw-public-schools-master-dataset>
>> into OpenStreetMap in line with the Import Guidelines
>> <https://wiki.openstreetmap.org/wiki/Import/Guidelines>.
>>
>> At this stage I'm seeking buy in from the local community as well as
>> feedback on my plan, before taking it to the imports list. Please refrain
>> from jumping the gun by importing this data before this review has been
>> completed.
>>
>> The data is CC BY 4.0 licensed (although the link above says only CC BY,
>> I've confirmed via email that it is CC BY 4.0) and the OSMF CC BY waiver
>> has been completed
>> <https://wiki.openstreetmap.org/wiki/File:CESE_NSW_AU_MasterSchools.pdf>.
>>
>> 1. Attribute Mapping
>>
>> Here is a sample of the attribute mapping I've applied, I'd appreciate
>> any feedback on this.
>>
>> amenity=school
>> access=private
>>     // Although generally on public land, access to public school grounds
>> is similar to any other private property, you can be there by invitation
>> and you can walk up to the front door but the owner/school can ask you to
>> leave at any time.
>> addr:city=Crows Nest
>>     // suburb
>> addr:postcode=2065
>> capacity=927
>>     // Based on enrolment numbers, although they aren't necessarily equal
>> I think in lieu of better information it serves as a good placeholder
>> contact:email=northsydbo-h.school at det.nsw.edu.au
>> website=http://www.northsydbo-h.schools.nsw.edu.au
>>     // derived from the email, but looks fine
>> contact:fax=+61 2 9957 6310
>> contact:phone=+61 2 9955 1565
>> fee=no
>>     Public schools don't have compulsory fees to attend, unlike most
>> private schools.
>> isced:level=2-3
>>     Upstream uses Infants, Primary, Secondary which are mapped to the
>> https://wiki.openstreetmap.org/wiki/Key:isced:level levels 0, 1, 2-3
>> respectively.
>> grades=7-12
>>     https://wiki.openstreetmap.org/wiki/Key:grades
>> name=North Sydney Boys High School
>> operator=NSW Department of Education
>>     // being part of the NSW Public Schools dataset implies they are
>> operated by the NSW Department of Education which in tern implies they are
>> public schools
>> ref:au.gov.nsw.cese=8132
>> ref:au.gov=7614
>>     // these references although not necessary might make it easier to
>> keep data in sync with future upstream updates
>> school:gender=male
>>   // mixed, male, female
>> school:selective=yes
>>   // yes, no, partial
>> school:specialty=comprehensive
>>   // agricultural, languages, arts, comprehensive (most)
>> start_date=1915-01-01
>>
>> A question is should we apply the source tag to the changeset or the
>> object. What should we do with the existing source tag? If it's obvious it
>> relates to the geometry I suggest it be moved to source:geometry, otherwise
>> I'd suggest it be deleted (it's still there in the history). Though a
>> specific source:name tag should be retained.
>>
>> I'm proposing to use on the changeset tag:
>>
>> source:url=https://data.cese.nsw.gov.au/data/dataset/nsw-
>> public-schools-master-dataset
>> source=NSW CESE Public Schools Master Dataset
>>
>> Along with a comment pointing to this thread.
>>
>> 2.  Import Plan
>>
>> To make importing the data easier, I've put together a basic web
>> application at https://andrewharvey.github.io/au-nsw-public-schools-to-
>> osm/diff.html. Based on matches identified by distance to nearest school
>> within 200m you can see which tags will change. It uses the JOSM Remote
>> Control to load the change into JOSM where the final upload(s) will take
>> place.
>>
>> My import plan is to go through this and and apply the changes or edit
>> them manually as necessary. In cases where tags conflict I plan to open
>> changeset comments to ask the author to determine what to do.
>>
>> I plan to use an dedicated imports account.
>>
>> I plan to add the attribution to the Contributors page.
>>
>> Most public primary schools have interchangeable names "Foo Primary
>> School" and "Foo Public School". If a different name is already tagged, I
>> propose we move it to alt_name, that makes the names consistent, but also
>> means that it might not match the name used on the ground. What do people
>> think about this?
>>
>> All the code and cached versions of the data files are available at
>> https://github.com/andrewharvey/au-nsw-public-schools-to-osm.
>>
>> Some stats...
>>
>> Total features from OSM: 2636 (1649 matched, 987 unmatched)
>> Total features from Upstream: 2209 (1713 matched, 496 unmatched)
>>
>> Of most interest are those 496 features from the upstream dataset not
>> found in OSM, but the other 1713 are still of interest as they add a lot of
>> missing tags to the existing objects in OSM. I haven't yet looked through
>> the "schools" we have in OSM but the dataset doesn't have, because the vast
>> majority will be private schools, and not things we need to investigate
>> further.
>>
>> I'm aware there are a number of "schools" in the upstream data we might
>> not consider schools for OSM for example "Field of Mars Environmental
>> Education Centre", "Royal National Park Environmental Education Centre" as
>> the students attending here are likely on excursion. However given the wiki
>> says "place where pupils, normally between the ages of about 6 and 18 are
>> taught under the supervision of teachers.", I think we should include them.
>>
>> There are a number of schools for the disadvantaged, disabled, etc. and
>> in hospital schools not currently mapped in OSM, importing this data means
>> we can include more of these kinds of schools. Unfortunately they aren't
>> tagged as such so they appear the same, but I still think it's better to
>> have them than not.
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-au/attachments/20180429/8e842114/attachment.html>


More information about the Talk-au mailing list