Hello! I wanted to discuss some ideas I’ve had for supporting better imports.
Right now, users who propose an import are expected to be quite technically inclined, and handle some arcane JOSM incantations in order to import their proposed data. There are some major pain points that I think could be handled with better tooling:
- Data auditing - one of the most important steps is to audit the quality of the data you’re importing. Without knowing if the data is good or not, it’s hard to get community buy-in. This is the part I’m currently stuck on in my own import proposal process, just because it’s a lot of manual work. I think it would be cool to have a website that allows you to publicly audit either against OSM or against another source (like the feature’s website).
- Conflation - currently this is best handled with the JOSM conflation plugin AFAIK. However, conflation is such an important part of the import that it should also be transparent. In addition to proposing the imported data, the proposed conflation should also be public and auditable.
- Dividing up the work - the guidelines basically say you need a goldilocks sized changeset - not too big… not too small… just right. This can also be difficult to do manually. If the data could be divided up for you into correctly sized changesets that would be great. This could also allow multiple users to help with the data, similar to how the OSM Tasking Manager works.
My experience has been with the HIFLD Hospitals dataset, which is basically just nodes. I think this is pretty much the simplest kind of import, so maybe the issues I’m hitting would be too difficult with other datasets.
Does anyone know of tools that currently exist to address these pain points? I’m interested in developing some better tooling, but I’m wondering if anyone else has these same problems.