Resources available now
- Find Suspicious OpenStreetMap Changesets (by number of objects created/modified/deleted, new contributers), see announcement
- new contributors feed - Personal feed of new contributers in an area. For covered areas see public BBoxs Last x days layers.
- osmcha.mapbox.com instance of OSMCha (OSM Changeset Analyzer) - Analyze and review changesets, suspicious changes are flagged
- detect_osm_weirdness.py - An OSM "weirdness" detector: Reads minutely/hourly diffs and looks for oddly-shaped ways or ways that don't quite make sense.
- osm-weirdness - a tool to find weird changesets
- OSMCha - Python package to detect suspicious OSM changesets
- OSM Hall Monitor - Suite of Python tools: reads diff files looking for large edits, skewed ratios of deletions to total edits or modifications to total edits, or changesets with modifications across the map; also can watch specific users or objects, with notification available; smart geospatial comparison coming soon (objects of certain shapes, etc.).
- OpenStreetMap Analytic Difference Engine.
Possible other resources
- Overpass API Augmented Diffs
- osc files
- osm files with different creation date
Possible libraries/algorithms to use
- Support vector machine (source)
- osm-stream-process - A simple Python script to make it easier to parse the minutely diff stream and perform tasks in near realtime.
- osm-watch - Advanced watching of OpenStreetMap changes (offline); Java, based on Augmented Diffs
- OSM Analytic Difference Engine, Diary entry - analytic live (web) service with text-based summary and visual-diff; Python, based on minutely diffs and main API
- Gabbar - Guarding OpenStreetMap from harmful edits using machine learning
- osm-changeset-classification - OpenStreetMap Changeset Classifier For Detecting SPAM, Imports, Reverts, and Mapping Errors
Important: none of the below criteria alone means anything. But combined it may be worth a look...
- new user
- user name changes versus id
- number of edits
- number of changesets
- area of changesets
- lots of special actions like
- edits spread over a wide area
- lots of deletions
- movement of data over a long distance
- high version numbers of objects (edit war)
- operating times (bots?)
- waylength above threshold
- way node reduction
- number of unique/similar changeset names
A periodically generated report could list possible
to be inspected closer. It could even introduce a scoring system.
- I'm dreaming of a tool that could warn you by email of any change made on a object you modified in the past which has defined properties.
- having the word "survey" in the note tag
- having a note, or a fixme, or a source tag
- in a zone of interest
- deleted object
- moved object
- Neis, P.; Goetz, M.; Zipf, A. Towards Automatic Vandalism Detection in OpenStreetMap. ISPRS Int. J. Geo-Inf. 2012, 1, 315-332.
- Nitasha Singla, Sukhjit Singh Sehra, Jaiteg Singh, A Review on Vandalism Detection in OpenStreetMap Data and Emerging Trends, American Journal of Networks and Communications. Vol. 3, No. 6, 2014, pp. 77-83. doi: 10.11648/j.ajnc.20140306.12
- Smart VGI Platforms: Methods for Automatic Vandalism Detection - Confirmation Seminar, Alireza Kashian - 13 March 2014
- Andrea Ballatore. "Defacing the Map: Cartographic Vandalism in the Digital Commons." The Cartographic Journal 2014; 51(3), 214-224. DOI: 10.1179/1743277414Y.0000000085
- Vandalism in Volunteered Geographic Information revisited - GIScience News Blog, Jan 2015
Examples of vandalism and mapping accidents:
- user blocks
- An open database of inconsistent edits observed on OSM from OSMCha
- German forum: Beispiele für Mapping-Unfälle und Vandalismus gesucht (Searching for examples of mapping accidents and vandalism)
- Editor issue leads to unexperienced users tagging underlying landuse instead of POI (examples in comments)
- How Afghan Amateur Mappers Unintentionally Punked Apple (using old OSM data)
- Somehow a urinating Android ended up in Google Maps
- Help fight advertising (Talk-us) - SEO Spam discussion + example list