Guideline for using linear data/linear attributes

General talk
1

Hi,

This is a question regarding both private and public (displayed in OSM) data. I guess answer might be different depending on the case private/OSM-public.

So the question is how do you display linear data/linear line attributes?
One common example max speed for a linear area.
More obscure/private examples could be road structural condition, or which maintenance team is in charge of cleaning/removing objects from a given road portion.
The data might be different depending on the side of the road including, obviously, single roadways.

The general idea I guess is that you just select a line and add relevant information.

Now lines and their nodes are typically defined by their own logic.

On the other hand linear information, such as speed limit might change in the middle of an existing line.

So the idea I had:

  1. Just carve the line objects so they match the information you have.
    I guess this might be acceptable for an proeminent data such as maxspeed but would be less acceptable for less common data.
    Of course this would not be acceptable at all for private data that is of value only to their owner.

  2. Next idea would be add line objects on top of existing line. I’m only a beginner* with OSM so I’m not sure if possible and how appropriate it would be regarding the OSM general philosophy.

  3. Third idea would be: define linear zones by a couple of points (or nodes) [beginning-end] + road name (/line name) and let the post processing do the work. This is robust, it is very operable with third party system, such as modern image recognition system that can pinpoint road signs. The downside is that it would require a lot more post processing to access the information.

So here are a bunch of idea off the top of my head.
But I image there are customary methods. Obviously I’m looking for the most standard method that would answer my need.

So if anyone can point me to relevant ressources.

Any help is welcome.

Thanks.

  • for the record, at the moment I’m trying to determine how appropriate OSM would be for my needs and if it is doable for me to migrate from proprietary standards.
2

For public data, you split the way.

For private data, there are no guidelines, as private data should not be in the OSM database, so how you handle it is purely a private matter.

I would, however, point out that additional splits can be added to the OSM way, nodes can move around slightly, or be added, and, unfortunately, people can delete and re-add, when they should be modifying.

Although you still need to cope with the delete and re-add case, I’d suggest you might want to aggregate a number of ways into a logical road, and then record the boundaries as percentages of the linear distance along the aggregated way.

3

Hi, thanks.
Of course private data shouldn’t be in OSM, nonetheless I imagined that OSM would somewhat promote description conventions-guidelines to data outside its direct perimeter.
My takeaway is that there is no easy way to implement this the current OSM data model, apart for public data where the guideline is to use splits, which is good to know.
Yes you could use aggregated distances but then anybody moving nodes might derail your recordings.
Anyway, good to know.

4

Yes, OSM + “permalinks” to the elements would rock! :smiley:
I don’t know if the effort needed to add this kind of indirection was evaluated…