Validation Scoop: Inconsistent edits observed

Posted by nammala on 26 May 2017 in English (English)

Resuming from previous diary posts, we would like to start off with interesting changesets we have observed over the last month. Nearly all of these changesets have been flagged by currently existing filters in OSMCha.

Here is what we found

image We found randomly added parks in the form of letters K N N in this changeset. This was detected while we were reviewing changesets from pokemon nest by new user filter in OSMCha.

image Added park over residential area. Here is the changeset. This was detected by park added by new user filter.

image We found a juggling teddy bear in Belarus which was posted in the Russian forum. Here is the associated changeset.

image We found a changeset which consisted of fictional edits that were basically some names and random characters. This was flagged through feature overlap filter. Feature overlap filter takes a feature and downloads nearby features to check if there is an overlap. We have been testing this filter with thresholds of overlap. Code for this can be found here.

image We found fantasy water bodies over a residential buildings. This changeset was detected by waterbody added by new user filter.

image Similar fantasy parks and waterbodies were found in this changeset. It was detected by waterbody added by new user filter.

image A park was found over buildings. This changeset was detected by park added by new user.

image We noticed suspicious edits in Japan by a new user, who is adding random parks, waterbodies, footways etc. This was detected by Park added by new user , Waterbody added by new user, and Footway added by new user filters in OSMCha.

If you are interested in getting into details of Pokemon specific edit detection, you can refer to the previous diary post for better understanding.

This is a very small subset of issues we have reviewed. Rarely do we find these kind of intentional errors on the map. We also have come across some detailed mapping activity. One of those is this changeset which added some beautiful details to a castle in Ludwigsburg. ✨ gif

Look forward to more such validation scoops from us. Let us know if you have feedback on any of these OSMCha filters. It would be great to hear similar stories of inconsistent or detailed mapping activity that you have come across. Comment on this diary post and share them with us (We ❤️ screenshots).

Comment from Zverik on 27 May 2017 at 11:35

It’s not KNN, but КИЙ: a Russian work for billiard cue / pool stick.

Awesome write-up, thanks for your work monitoring the edits and post more screenshots! :)

Comment from Urbán Máté on 29 May 2017 at 23:54

Pretty rough edits in Russia. I haven’t come across similar things in Hungary, yet.

Comment from IamColleen on 5 June 2017 at 04:02

Thanks for this - I think this would be a very cheeky way to show organizations who are skeptical about OSM data (validation) how inaccurate edits are flagged; that they are.

Comment from gorn on 13 June 2017 at 07:44

@IamColleen assuming optimisticaly that it catches all/most of vandalism

Comment from Saikat Maiti on 4 October 2017 at 20:59

Thanks @nammala. This is really important thing to know.

Login to leave a comment