The mapping team at Meta is working to improve the data quality in OpenStreetMap (OSM), the free wiki world map built by a community of mappers that contribute and maintain data all over the world. Our goal is to collaborate with the OSM community to create quality data validated by trained mappers and local OSM communities. We perform various editing projects such as AI-Assisted Road Tracing, vandalism and profanity cleanup, and MapRoulette Challenges. Please check out Mapwith.ai and our Meta OSM Data Team GitHub page for more information about our projects. For any other questions, please send an email to osm@fb.com
Map editing contributions
AI-Assisted Road Tracing
Since 2017, Meta has been running a project to perform AI-Assisted Road Tracing within OpenStreetMap. The project has included road tracing in Thailand, Indonesia, Malaysia, Vietnam, India, and Tanzania.
Multipolygon Relations Cleanup
Multipolygons are a type of relation that are vital for displaying complex geometry. These relations are often unintentionally broken in the course of mapping which can lead to prominent features failing to render on the map. Our team gathers a list of multipolygon relations that cannot currently render and perform fixes using JOSM. Relations cleanup edits are given the following hashtag: #RelationsCleanup
This project uses a Meta internal profanity detection model to identify profanity on name labels. A human review process is conducted on detected features to determine if the name is in fact profanity. Submission of name label fixes are performed based on manual review results. Profanity Cleanup edits are given the following hashtag: #ProfanityCleanup
This project uses an internal detection model to identify vandalism candidates. A human review process is conducted on detected features to determine if the feature has in fact been vandalized. Submission of vandalized feature fixes are performed based on manual review results. Vandalism Cleanup edits are given the following hashtag: #VandalismCleanup
Our team reviews numerous edits by using OSMCha and osm-compare detection rules to help improve OSM data quality. Mappers analyze the flagged features and determine if a fix is needed. Critical changesets flagged by mappers within the community are also reviewed. OSMCha edits are given the following hashtag: #OSMCha
Here are some of the issues/flagged reasons we commonly focus on:
large-building
Dragged highway_waterway
pokemon_edits
MapRoulette
Our team contributes to various MapRoulette challenges to help improve OpenStreetMap data.
Atlas Checks
We use Atlas Checks to validate potential map data issues in specific regions. Changesets contributed to OSM through Atlas Checks detections are given the following hashtag: #AtlasChecks.
For an up-to-date list and explanation of individual checks, please visit the Atlas Check GitHub.
Other Data Fixes
Alongside the major data fix processes above, we submit changesets to address data issues as well for various review processes. Changesets submitted for data fixes and don't fall into the major processes above, are given the following hashtag: #dataupdate
Building Mapping
Our team is mapping buildings either using Esri/ArcGIS Datasets available in RapiD or by manual mapping. Building edits are given the following hashtag: #buildingmapping
Pedestrian mapping
Meta is working with local communities to map pedestrian infrastructure. This includes the mapping of sidewalk ways and crosswalks. See our dedicated pedestrian mapping wiki for more details.
Editing Guidelines
The mapping team at Meta edits OSM by following the rules and guidelines that apply to the OpenStreetMap community. Local knowledge and communication are paramount. When in doubt, our mappers will leave a changeset comment and post a question on communication channels such as telegram, Slack, or in the OSM forums, if needed. For any feedback and questions, please send an email to osm@fb.com
Edits in China without consulting local community, even destroy correct data (such as delete part of name, or change Simplified to Traditional) and describe as a "name restored":