Severin I like the ideas here. Essentially, we're talking about calculating and recording mappers "reputation", and then incorporating into handling of validation.
"How did you contribute" http://hdyc.neis-one.org/ is one example of reputation calculation. It's entirely automated. Your suggesting to add specific, manually recorded feedback on a users specific edits. A prerequisite would be to categorize OSMTM taks by the general kind of work required, which could have other benefits (like standardizing instructions). I wonder if we could have the same result, to focus validator efforts and give feedback to mappers, by simply using the HDYC reputation calculation ("Type of Mapper"). Are most cases of mistakes in HOT jobs from relatively new mappers to OSM? Or do we also see experienced OSM-ers making mistakes in unfamiliar terrain? In any case, the work on the OSMTM2 API could help. We can then more easily analyse active jobs, by querying and maybe even setting status through the API. Another idea I've been mulling over is "microtasking" validation. The idea is to make in depth validation easier; it seems that validation is not so popular or easy with our current tools. Split up the validation, of say buildings over a task square, into individual building microtasks. A contributor simply marks whether the building was drawn correctly, or not. This should only take a few seconds, and could involve large numbers of people in a very simple task. The collective microtask results of a task square could then be used to guide selective in depth validation. -Mikel On Sunday, August 31, 2014 3:09 AM, Severin Menard <[email protected]> wrote: > Hi all, > > Thinking aloud, would not be possible (actually asking tech people) to have a tool allwing the > following: > - detecting changesets with hotosm hashtags and picking up the username > - comparing the username to a list of HOTOSM contributors and stating if it is new or not and > already validated for a certain numbers of quality items (like drawing buildings correctly, drawing > roads correctly, tagging roads correctly, etc.) > - when contributors have not been validated yet, a task is sent to a validation team > - one team member picks up the task, check, validate the work of the contributor and contact > her/him if some mistakes. A form would be great, with checkboxes for typical errors, and if > making a typical one, the contributor would receive in the answer link to the Learning point (there > are already quite a few in LearnOSM) dedicated to this error > - once done the hotosm contributor quality status for this contributor would be: good for such and > such aspects, bad for such and such ones and the latter would then be tasked in the future for > validators as soon as this contributor would submit a new changeset > > Thoughts? > > Sincerely, > > Severin
_______________________________________________ HOT mailing list [email protected] https://lists.openstreetmap.org/listinfo/hot
