Pin points versus Areas choice for businesses?

I often see Areas and pin points being confused. For example, ‘bakery’ would be selected as the feature type of an area. And other times, a pin point would be used for the bakery on a building.

My understanding is that if a business is on ground level of a building, and on the other floors it is used for other purposes (offices/apartments…), I would use a pin point. And when a building is only used for the business’ purpose (like a big supermarket or a swimming pool…), I would use areas. But maybe I’m wrong?

Does anyone know the rule for when to use pin points and when to use areas for businesses/services?
Should I correct the ones that got it confused?

Thank you! :slight_smile:

There are conflicting views on that. Some feel you should tag the building as a building then superimpose another area as the business. I see two problems with that:

  1. In the iD editor, if you totally superimpose two areas it is not (currently) possible to choose which to access. So if you need to change some detail about the building you may find you can only edit the business, so have to delete the business in order to change the building then add the business again.

  2. Obviously, the building has an address. But it is also desirable to give the business an address, for people who use the query functionality. Having the same data in two places it not really a good idea.

Others disagree (at times, quite strongly) with me on that and insist you should map the building and business separately despite the iD editor making it seductively easy to combine the two things in a single map object and easy to amend details later.

It is a trade-off. Sometimes you’re in a hurry because you have a lot of detail to add in a large area, or the aerial imagery isn’t clear, so you add a point. Ideally, though, add an area because more detail is usually more useful.

If there’s a lot of stuff in the area that hasn’t been mapped at all, it’s better to concentrate on mapping that first. If the area is pretty complete then feel free to tidy up.

My own view on these things is that we’ll never get it 100% right. But if we can improve it, and have the time, then it’s worth doing. The only real criterion is “Which of these possible actions I can perform makes the map most useful to consumers?” Other people may have different views.

Generally, anything is better than nothing, but more is better.

If a shopping street is extensively mapped, the first thing to concentrate on is probably dealing with all the changes of tenants and brand names. People can go in, create a detailed map, then abandon the area, because it is more exciting to add new than to maintain.

I do think that it is wrong to treat the whole building as a single business when the business only occupies one or two floors.

There are actually two separate decisions here which are often mixed up:

  1. Should the object be mapped as a point or as an area?
  2. Should the tags for a shop or other occupant be added to the building itself, or a separate element? (Keep in mind that you can draw an area for a shop itself, using an area does not necessarily imply re-using the building’s area.)

As for question 1: Areas are strictly better than points, that’s pretty much unambiguous in my opinion. So upgrading existing points to areas is always ok. But of course, points get you most of the usefulness with only a fraction of the effort, so they are often the most sensible first step for mapping.

Question 2 is less straightforward. When there’s more than one “point of interest” in the same building, it’s pretty clear that separate elements should be used. But apart from that, opinions are mixed. I think it’s relatively accepted (but not a 100% consensus) to add shop tags to a building if the entire building is occupied by that shop. Due to the different opinions on this issue, I strongly recommend seeking an agreement with the other mappers in your area before “correcting” things.

Not so many years ago, when the quality of aerial imagery was much poorer, points were about all you could do with any confidence.

When I started mapping the town where I live, about 7 months ago, most of the businesses that were mapped had been mapped as points about 3 years ago. Some with positional errors of up to 30 metres, and on the wrong side of the road. Back then it was about the best that could be done (especially as it seems that it was done by armchair mappers with no local knowledge).

So mapping as points may not have been a decision made to save time but possibly the best that could be achieved at the time.

Even when the shop occupies the complete building there are tags that are applicable to the store and to the building as well, e.g. start_date. How do you record both dates (e.g. newly opened restaurant in historic building), when you only have 1 area ?

Yes, I should have mentioned this as another factor. In situations where such tags are present (other examples: building has a name by itself, building has a Wikipedia link that does not apply to the shop, …), separating the two is the best choice.

Personally, I’d even be happy if we always separated them, as it makes more sense semantically. But that’s does not (yet?) seem to be common practice, so I was trying to describe what I perceive as the status quo.

In The Netherlands we imported all buildings and addresses from government data over the past few years. In this process we came to the conclusion that in our case it was best to separate addresses/poi’s from buildings. This also proves to be of use in the maintenance process of new/demolished buildings and addresses.

I’d rather not represent POIs such as school, universities, manufacturer plants ,etc as points. And I want to see the address on those amenities, leisures, … areas

And what about pubs consisting of inside and outside spaces (terraces, playgrounds, gardens, parkings) ? I want to represent them as areas, not as points.

I’m not (yet) into indoor mapping of malls, but in indoor mapping one uses areas for shops, not nodes.

Addresses are IMHO attributes, not objects on their own. They can go on points or areas, whatever is most appropriate for mapping now. But in the end they always apply to an area, not a point.

So for me, nodes for POIs are a temporary solution. We will evolve to a model where POIs are represented by areas. Areas that have to be drawn separately from buildings, because the POI has different attributes from the buildings, because they occupy a part, have a different address, or the building is only a part of the POI, or we have detailed indoor mapping of the POI.

I agree. IMHO an address is always associated with a plot, or some building on that plot.
Also, as addresses are also associated to entrance point, large buildings or plots could have two or more addresses, with different addr:housenumber or even different addr:street.