How to detect suspicious OpenStreetMap Changesets with incorrect edits? – Neis One!
How to detect suspicious OpenStreetMap Changesets with incorrect edits?
Since its rise
in popularity, the well-known online encyclopedia Wikipedia has been struggling with manipulation or, in the worst-case, vandalism attempts. Similarly, the
OpenStreetMap
(OSM) project suffered several times over the past few years of cases where incorrect map data edits were made. These erroneous edits can stem at times from (new) contributors or illegal data imports (or automated edits) which have not been discussed in advance with the community or the
Data Working Group
DWG
) and corrupted existing project data. The current
OSM wiki page
gives a great overview about general guidelines and e.g. types of vandalism. Another page in the wiki also mentions a prototype of a rule based system for the automatic detection of vandalism in OSM, which I developed in
2012
. However, the system has never actually been implemented.
Today
, the contributors of OSM can use a variety of different tools to inspect an area or particular map changes. A few of them are listed below (complete list can be found
here
):
WhoDidIt
ITO world OSM Mapper
Last Week on OSM
achavi – Augmented OSM Change Viewer
Based on the database
which I use for multiple other services, I created an easy to use webpage to find suspicious OSM changesets with possibly incorrect map edits. The webpage offers some filter options such as the boundary of a country or the object change of interest. In contrast to the other aforementioned webpages you can also filter changesets based on the active “mapping days” of the contributor. A “mapping day” is a day on which the contributor created at least one changeset, independent from the registration date. I am also planning on adding additional user reputation information such as used editors or tagging behavior. And of course I am going to add some RSS feeds in the next version. The first version can be found
here
What makes all of this different from other tools?
Well, I think one of the major advantages is the simplicity of the webpage and that you can filter changesets based on the contributor activity and/or the changeset edits. In contrast to other tools, you can find changesets not only based on your area of interest, but also based on potential beginner mistakes and hopefully not vandalism attempts or fictional/ none existing map data.
Find Suspicious OSM Changesets here:
Thanks to maɪˈæmɪ Dennis.
Posted
January 20th, 2016
in
Allgemein
Map
OpenStreetMap
Stats
by
Pascal Neis
Tags:
ChangeSets
Contributors
OSM
Planet
Quality
Stats
Website
World
Comments
16 responses to “How to detect suspicious OpenStreetMap Changesets with incorrect edits?”
naoliv
January 21st, 2016
\o/
January 21st, 2016
Great work! This kind of information is crucial for the continuing success of the project as it attracts new contributors. Also useful would be the ability to flag changesets for review.
Marián Kyral
January 21st, 2016
HI. Really usefull. Could you also put link to achavi? It shows what changed much better.
pascal
January 21st, 2016
Hi Marián, great idea. Just added.
Marián Kyral
January 22nd, 2016
Thanks. And what about link in popup? 😉
pascal
January 22nd, 2016
sure 😉
Marián Kyral
January 26th, 2016
Excelent. Thanks
Miroslav Suchy
January 21st, 2016
Great! Can you enhance it that in select box with deleted/created/modified will be “any” and the same for nodes/ways/relations? So I can review work of any newbie?
pascal
February 13th, 2016
Hi Miroslav, also good idea. Added.
DaCor
January 22nd, 2016
Nice work as always Pascal, very useful
alphensebezorger
June 12th, 2016
Very useful tool. It doesn’t show the changeset comment anymore unfortunately. Am I doing something wrong, or is this part of the tool broken?
pascal
June 12th, 2016
Thank you very much for noticing, there was something broken. Has been fixed.
alphensebezorger
June 12th, 2016
That was quick! Thank you for the lightning fast repair.
Kevin Kenny
June 29th, 2016
Nice job! This will be helpful for detecting new bots that aren’t playing by the rules. When I open up to ‘last 24 hours’ or ‘last 48 hours’, my bot certainly shows up! (It’s been discussed on imports and talk-us, and there’s a page for it on the wiki, linked from import/catalogue. I don’t expect your program to be able to tell that, though.)
DF
December 20th, 2016
Nice tool. It would be cool to see edit frequency for a user across multiple changesets. So for example if someone makes 1000 changsets in a day with 10 nodes each, it will be just as obvious as 10000 nodes in 1 changeset.
Filtering by node/way creation/deletion rate would also be nice. Thus it would would be easy to spot obvious automated mass edits.
pascal
December 24th, 2016
Thanks for your comment. I’m working on a new version, which also has some new features. Sadly, it needs time …