since my first comment some things have transpired that I am less on board with. The following is me as @osmf-data-wg member:
As others like @Minh_Nguyen have said this does buck the norm for this sort of POI editing. It has been status quo in many regions to map strip malls as one or two single large buildings are visible from aerial. It is very difficult to segment a single building up if the stall widths aren’t precisely known and visible on aerial. Believe me–I’ve tried this personally using the air conditioning equipment and ground survey with EveryDoor. It is quite challenging to get right and adjusting the buildings (once split) is hard than moving a node.
My colleague @Stereo laid this point out precisely. More time is needed to review.
The source of this edits is the plan of the quarter. It is shown that object as separate buildings with different house numbers. Moreover, I used some photos (like this) to recognize objects. One more case - there are ledges on the roof which are possible to use as markers on the satellite imagery.
I’m apologizing that I didn’t put the links into the comment.
Could you please guide us on the approximate time frame for making a decision on the part of the community, so that we can understand whether we will be able to complete the work by the deadlines agreed with our customer.
Per the organized editing guidelines:
"This should be done no less than two weeks before the activity is started. "
That doesn’t sound very urgent to me TBH. Also, it is not clear that having units a separate buildings provides better navigation as opposed to just POI nodes.
For the building north of Hansell Street Northeast and west of Church Street the “plan” you cite shows seven units, but after your edit, there are five buildings.
But in this case, the ledges do not support the edits you made, at least not fully.
It would be very helpful if your sources were cited in the future.
In general, I really have no idea what you are planning on doing as you have not explained the project in sufficient detail. Just one example, are you moving tags from nodes to the new buildings, if so, which ones, and are you then removing them from the nodes so there are no duplicates (you created duplicates in the case we are discussing here)
For the building north of Hansell Street Northeast and west of Church Street the “plan” you cite shows seven units, but after your edit, there are five buildings.
The reason is the yoga center which occupies 3 parts.
Just one example, are you moving tags from nodes to the new buildings, if so, which ones, and are you then removing them from the nodes so there are no duplicates (you created duplicates in the case we are discussing here)
We are going to use tag building=yes for polygons and we are not going to copy information from points. (or building=retail for stores, if this tag is more relevant)
I understand my mistake with duplicated tags and can delete them from polygons. This was the only case.
This isn’t a good look for Mapbox. Could you please check with whoever normally manages the OSM community relationship there to see if there are standard internal guidelines for you to follow, or someone who’s done this before who can offer advice? Mapbox is risking potential consequences that are totally preventable.
Their wiki page contributors list hasn’t been updated to include Zhanna and Alexandra as well as the possibility of others.
Blockquote
We need to collect a ground truth dataset to evaluate the performance of our algorithms
Why is osm data, let alone production data and servers, being used to test an algorithm? This goes against any type of best practice. Why is this not being done on a dev server? Why did you choose partially tagged buildings(your own words)? Why are you trying to fit this in to days when the process time is two weeks? Why not wait for local mappers that may know more about the area respond? What are the 70 other strip malls? What and how will this change affect OSM and our data integrity.
@ElliottPlack could we pull the change back for two weeks minimum to allow for discussion. At this point it looks like we need pictures in the wiki or changes to the definition of a strip mall, conversations about POI(I have already run into problems with a strip mall), and including definition and examples of common/shared wall buildings/stores which to some who don’t know may look like a strip mall.
@ZhannaB The data you give for reference isn’t enough to show or justify a change or even that the yoga studio to take up multiple spots. All I can gather from the info you have shared is that there are multiple store fronts for lease which wouldn’t jive with the claim the yoga studio having taken down walls probably done structural work to the building to take up multiple stores. This leads me to have concerns about what you are doing and the other changes you want to make as well as mapbox oversight of workers, quality of work, and standards/procedures, as well as vetting client requests for changes that don’t align with osm.
This is not our concern and should not be brought up. It comes off as an attempt to dodge the Organized Editing Guidelines, the need for which has been well demonstrated here. Perhaps we should discuss with the person making these undeliverable promises to customers.
@ZhannaB – Mikel here, I helped support OSM work for years when I was at Mapbox. Been involved with OSMF for years. These days I’m the Advisory Board Coordinator for the OSMF, lending a helping hand to organizations working in OSM. I think you can read here some genuine concerns you will need to grapple with. It doesn’t need to be slow, but it does need to be responsive. Mapbox has a representative to the OSMF Advisory Board, and I think they could be valuable to helping orient this discussion successfully. Are you able to loop in your colleague, or want me to?
@ZhannaB (and other Mapbox folks who might have an interest). I think you could help your cause if you could explain in more detail what you intend to do. Below is an example. I don’t know if the community would support this, and I am not even sure if it is what you intend to do since you really haven’t provided much detail, but it should give you an idea of the level of detail that would allow us to better review your proposal and provide feedback. My comments are in
Mapbox has a customer who is interested in multi-tenant retail complexes with external access (not covered malls that have internal access) [the features you have edited so far are not “strip malls” as many Americans understand the term].
This customer would like to see each retail business (e.g. shop=*) in its own building.
Mapbox has selected 70 cases where a single building=* contains more than one node that is a retail business, you can access the current data for those 70 here: [provide link so the community can review] These are the features we will be editing as part of this project. This project is a test, and we may eventually be editing many more features, but we will seek community feedback before proceeding further.
Mapbox will be splitting the existing buildings into separate, but connected (i.e. shared nodes) buildings representing the area occupied by each business.
Tags for the new buildings will be copied from the existing building unless we come across evidence that they are obviously wrong. Mapbox will not be moving tags from nodes representing businesses to the buildings.
Nodes representing businesses will be left as is, and their tags will not be transferred to the new buildings. However, if through our research Mapbox discovers a missing business, or a business that no longer exists, Mapbox will correct the OSM node data accordingly. In making these changes Mapbox will take into account any edits, especially recent edits from “survey”, by local mappers.
If a single business occupies more than one contiguous storefront/suite its total area will be represented by one building.
In addition to the commonly used OSM imagery sources, Mapbox will be doing web research to find other reliable sources with suitable licenses to support our changes.
For each changeset Mapbox will cite the sources used, and only those used, to make the changes in that changeset [too often mappers have a list of 10+ different sources that they attach to every changeset regardless of whether they support the changes made in that changeset, this makes community review more difficult as one has to pick through a bunch of sources that do not support the edit in order to find those that do]
Hi folks. Apologies for coming to this thread late. @ZhannaB and I will be reviewing the plan for this work and our policies for communicating about mapping projects going forward.
@Tom_Lee Thanks! I look forward to seeing a proposal with enough detail that the community can evaluate what Mapbox is proposing to do here.
I am not saying I agree with this, but this is what I think Mapbox intends to do based on the couple of edits that were already made. Hopefully, we can get some clarity on the matter (and other details of the proposed organized editing) so the community can weigh in on whether this is an acceptable approach.
@tekim we’re still discussing internally, but I do want to be up front that one possible outcome is this proposal not being pursued: if the OSM practice for strip malls doesn’t match the kind of representation we need in the short term, we might wind up pursuing a different approach that doesn’t involve OSM edits at all. I don’t yet know enough to say if that’ll be the case, but I did want to be sure you aren’t left hanging.
What I can promise is that if we do proceed with mapping these locations, we’ll get you that fuller sense of clarity after doing our best to align our plans with the community’s norms around mapping strip malls.
Once Mapbox clarifies what it intends to do I think the community can more fully comment. At this point, we don’t know. Thanks again for stepping in on this issue.
