Slide 1
Slide 1 text
Validating the map
State of the Map 2017, Japan
user:PlaneMad | Arun Ganesh
Slide 2
Slide 2 text
Crowdsourced data projects
Slide 3
Slide 3 text
Cities can become shops
Slide 4
Slide 4 text
Or an office
Slide 5
Slide 5 text
Or just get a bad name
Slide 6
Slide 6 text
Why does it happen?
Slide 7
Slide 7 text
My mouse moved
Slide 8
Slide 8 text
It’s time to rename my city
Slide 9
Slide 9 text
This looks prettier
Slide 10
Slide 10 text
The virus has caught on
Slide 11
Slide 11 text
This description tag looks useful
Slide 12
Slide 12 text
Need PokeStop now
Slide 13
Slide 13 text
This is a nice place to draw
Slide 14
Slide 14 text
From where?
Contributors
or
Software
Slide 15
Slide 15 text
60% of harmful changes
are from
New contributors
Source: OSMCha
Slide 16
Slide 16 text
New contributors
Naturally make mistakes
Not aware of OpenStreetMap
Not aware of contribution guidelines
Not familiar with editing tools
Not familiar with tagging system
Not aware of community
Slide 17
Slide 17 text
Experienced contributors
Make large scale edits
Use powerful tools
Edit important and very complex features
Not scared of making a mistake
But mistakes happen
Slide 18
Slide 18 text
More new and
active contributors
than ever before
Slide 19
Slide 19 text
Software
Simple tools attracts new users
Power tools allows large scale editing
Don’t give powerful tools to new users
Slide 20
Slide 20 text
Everything breaks
Slide 21
Slide 21 text
Protecting the map
No single strategy can be foolproof
We need as many eyes on the data as possible
Slide 22
Slide 22 text
We need to see the data, not the map
Slide 23
Slide 23 text
Foolproof data validation requires local context
Slide 24
Slide 24 text
We need accessible tools
for community
to peer review data
Over 30 tools listed on OSM Wiki
None on the OSM Homepage
Slide 25
Slide 25 text
30,000 changesets
by 4,000 users
everyday.
2000 new users/day
Slide 26
Slide 26 text
OSMChangesetAnalyzer
Slide 27
Slide 27 text
Filter changesets by location, date, category..
Slide 28
Slide 28 text
Visualize a changeset like Achavi
Slide 29
Slide 29 text
Peer review and tag changesets
Slide 30
Slide 30 text
103,680 changesets reviewed
14,133 found with bad quality
Use for machine learning?
Slide 31
Slide 31 text
Contribute
Source code https://github.com/mapbox/osmcha-frontend/issues
Write detectors https://github.com/mapbox/osm-compare