Hi All,

I tried to send the following idea to the "imports" mailing list the other
day, but turns out I wasn't registered to post there. I guess it fits well
with Summer of Code so I will add it to the wiki, but not being a developer
I can't mentor anyone.

----
<snip>

So my question is: How do we improve our import work flows so that it is
easier to keep imported or merged data up to date?

Some ideas:

1. Tools to compare OSM data against the available external data set. One
recent blog post on this by SK53
http://sk53-osm.blogspot.co.uk/2014/02/looking-for-footpaths-in-hickling-notts.html

2. We need to be asking for not just full datasets but also regular
change-sets. If we cannot get a change-set from the data supplier then we
need to be keeping a copy of any imported data and creating comparison
tools between the original imported data and the updated data from that
government/organisation.

3. More data conflation tools. Not all data can be imported in bulk. We
need to look at developing more tools to allow for piecemeal imports from
the local community. For example the Android app Vespucci *could* be
extended/forked to allow the following work flow:

3a. A new "import" dataset is added to a holding database.
3b. On the ground mappers can can then view this database on the ground
using Vespucci on their tablet/phone
3c. For each element they mark it as "verified" or "incorrect", and if
necessary change the tags or geometry using the tools already built into
Vespucci. At this stage Vespucci has not shown any other OSM data, just the
holding database layer. This makes it easier to use in areas of high data
density.
3d. If the data is just nodes then Vespucci searches for potential matches
in the OSM data. Is one is found the user is asked how to merge the two. If
not found then the node is imported into OSM.
3e. For ways the user can either work with Vespucci to merge/import the
data, or they can log in when back home using JOSM or ID and work with
these editing tools to merge the verified data from the holding database.

Any other ideas?

RobJN
_______________________________________________
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Reply via email to