Open data and mapping have enabled a large number of people to contribute using crowdsourcing techniques. Wide areas and a lot of detail have become much easier to map, while the data are amenable to different purposes for users. However, by creating open map data and user possibilities for mapping, application are also susceptible to map vandalism and general disruption by maliciously user intentions.
What is Map Vandalism?
Map vandalism is the deliberate wrong change or update to map data, such as adding an incorrect name for a street or place. Open Street Map (OSM), the most widely known open map data project, is potentially vulnerable to map vandalism.
Typically, data are monitored through a pipeline related to an edit or map submissions. For instance, Mapbox will perform automated pipeline checks that incorporate checks for vandalism and usually at least one human checker also looks over the data. However, even then, acts of map vandalism get through and can be posted, such as a well known event known from a few years ago.
While organizations have chosen OSM as the most common open map platform because of its wide appeal and user contributions, these same organizations are also well aware of potential vandalism. Using analytical models and machine learning techniques, organizations such as Facebook are attempting to detect acts of vandalism and flag these for edits.
OSM has created clear policy to also discourage vandalism and has provided examples, such as a fake town created with fake streets to demonstrate acts of vandalism. In fact, OSM has created policy such that simple acts of vandalism can be quickly reverted to a previous map setting and then edited if needed. More malicious intent may lead to a permanent ban for the users creating the vandalism.
How Prevalent is Map Vandalism?
In relatively recent research that looked at OSM in the popular Pokemon GO (PGO) game, it was found that vandalism can be common, but mostly fixed quickly. In fact, one benefit of platforms like OSM is that a large community means that vandalism is often spotted quickly.
Usually, acts of vandalism are fixed within minutes or hours, and are generally only disruptive sporadically. It was found that only 16.5% acts of vandalism persisted up to a week. Fixes are mostly pursued by a relatively smaller but dedicated group who take vandalism seriously.
Many acts of vandalism are somewhat localized, where vandals are located within the country where the vandalism occurs or even within a limited radius around the town they may come from. This implies that vandals might be best traced and monitored for their activity, where such individuals, if they vandalise again, are likely to do it in their local area.
The community monitoring and responding to vandalism also appears to be more efficient if they become aware that vandalism is on the increase, where users were found to pay closer attention to map details when there was greater awareness of vandalism.
Strategizes to Fix and Prevent Map Vandalism
One recent study showed that training machine learning approaches on local area vandalism, that is areas where the machine learning training data are gathered from, best work in local areas. In other words, as local vandalism is likely to occur from the same group of users, then the patterns of these vandalism acts tend to be similar across time in the same area.
Methods such as random forest techniques work well at detecting local acts of vandalism on maps, but when the training data are then used for more distant regions the detection of vandalism proved harder. In other words, to prevent acts of map vandalism, it might be necessary to use a variety of locally trained data with the trained models then applied for limited geographic regions.
Map vandalism is certainly a problem. Checks to prevent it are present and researchers have created automated techniques to detect it while also noticing how user groups respond to acts of vandalism that help prevent future attacks. Nevertheless, with open data becoming more common, we can expect map vandalism to be a persistent problem.
Researchers are now attempting to create better ways to find vandalism. Given that there are many individuals involved in vandalism, this may not be always successful. Nevertheless, the best form of vigilance appears to be the community of map users on platforms such as OSM. Having many individuals altered to the presence of vandalism often seems to be the best way to prevent or quickly edit acts of map vandalism.
 For an example of map vandalism that was well publicized, see: https://blog.mapbox.com/zero-tolerance-for-hate-speech-46293a18bba9.
 For more on map vandalism and Facebook’s attempts to catch vandalism, see: https://daylightmap.org/2021/05/24/name-vandalism-corpus-release.html.
 For more on vandalism on OSM and patterns associated with it, including how it can be prevented, see: Juhász L, Novack T, Hochmair H, Qiao S. Cartographic Vandalism in the Era of Location-Based Games—The Case of OpenStreetMap and Pokémon GO. IJGI. 2020 Mar 26;9(4):197.
 For more machine learning methods for vandalism detection, see: Truong Q, Touya G, Runz C. OSMWatchman: Learning How to Detect Vandalized Contributions in OSM Using a Random Forest Classifier. IJGI. 2020 Aug 22;9(9):504.