Just a couple of comments about data quality, a week ago in Ethiopia there
were 4,500 untagged ways or untagged with a comment, the most common is
area=yes. These do not show up in the map so a user in the field wouldn't
even know of their existence.
To me this a data quality issue. Things have been mapped but are not
appearing in the map.
Secondly talking to an American data analyst his issue was sometimes the
maps could be trusted and sometimes they weren't so good. In Nepal the
amount of effort needed to clean up the map was considerable. I seem to
recall Kathmandu Living Labs has been involved in a fair amount of work and
I think it was because they saw the value in cleaner data.
I don't think Ethiopia is exceptional.
On 13 October 2016 at 11:43, Mike Thompson <miketh...@gmail.com> wrote:
> Two situations that a new mapper may face:
> 1) Task area with nothing, or very little, mapped.
> This *should* be pretty easy. For example, in the case of buildings: draw
> building, square it, tag it as a building, don't draw two buildings over
> lapping each other. However, despite verbal instruction, demonstrations,
> and written instructions, many get it wrong. My conclusion is human
> learning is more complex than many of us assume (hence the need for
> professional educators in our society). We may need to adjust our training
> (both online and in person) to be more effective. Just providing the
> information is not enough.
> 2) Task area has already been partially mapped, either outside of the
> tasking manager, as part of a previous project, or as part of the current
> project where another mapper didn't finish.
> Essentially we are asking the mapper in question to be a validator,
> because we expect that when he or she marks the task as done, they are
> saying that all of the mapping is done according to the instructions, not
> just the mapping he or she did. This can be very difficult for a new
> mapper, especially without the tools in JOSM. For example, if the
> instructions call for buildings to be square, they have to pan the entire
> task looking for buildings where is in JOSM, I put all buildings into the
> to do list and step through them (I could also search for ways tagged as
> buildings with less than a certain number of nodes and square them all at
> * Have the tasking manager require a mapper to open (no way of knowing if
> they actually read them) their messages before they can do checkout
> another tasks.
> * Have tasking manager forward messages to mapper's email account.
> * Require that a mapper at least visits the "instructions" tab before
> checking out a task.
> * Enhance project instructions to cover "edge cases". For example, should
> buildings under construction be mapped? Should existing features be
> spatially adjusted to the preferred imagery source, or should the imagery
> be adjusted to them?
> * Enhance project instructions to have links to examples (some of the
> Africa projects already do this). "Not sure what a building under
> construction looks like? Click here for some examples." "Not sure how to
> adjust imagery offset in iD? Click here for instructions."
> * Mapathons should not be held without experienced mappers present who are
> willing to spend the entire event going from person to person to check on
> their work and answer any questions. We need many different people to make
> this work, and if someone just wants to be an organizer that is great, but
> they should ensure experienced mappers are in attendance.
> * Some sort of online way to quickly give feedback as soon as a save is
> made such that it is not necessary to wait until a task is marked as done.
HOT mailing list