Recent diary entries
State of the Map 2017 attendees
State of the Map 2017 was a lively event. This was my first attendance at a State of the map conference. There was good participation from experienced community members from different parts of the world along with members from OSMF, DWG and from companies such as Mapillary, Telenav, Facebook, Apple, GeoRepublic, Grab and more.
It is wonderful to be part of this vibrant community. I have Mapbox to thank, for introducing me to OpenStreetMap and allowing me to work with passionate people for mapping, documentation and validation. I would like to use this diary post to answer some of the recurring questions on OSMCha and some notes from State of the Map 2017.
I felt that there was a common agreement on peer reviewing amongst local mappers from the SoTM attendees. This promotes constructive feedback, builds a stronger community and results in better data quality.
It was great to meet Frederick Ramm from DWG. The workflow the validation team at Mapbox follows has been similar to how DWG approaches in fixing issues, escalations and in sharing feedback to users to become core mappers.
OSMCha is getting adopted amongst the community rapidly. Last week, 311 changesets were reviewed by various members of the community. The recent OSM feature to request a changeset review can help new contributors to map with more confidence and build a error free map. Do read Pascal Niess’ nice blog post about this and for other ways to review changesets.
If you are a developer or a team who intends to use OSMCha but if you prefer your own instance, feel free to deploy your own. Here are the resources
Documentation on using OSMCha is at https://osmcha.mapbox.com/about. It would be great to have this translated into local languages by community members. This will help OSMCha serve and reach out to local communities better.
- OSM-Compare is the open library of GeoJSON comparison rules that flags problematic edits in OSMCha https://github.com/mapbox/osm-compare
- We can use OSM-Compare not only for identifying odd changes on OSM, but also as a social tool or watchlist of certain feature edits:
- I saw passionate rail mappers at SoTM, I think it would help to flag changesets with rail edits so OSM users can identify and interact with each other, in local communities.
- Blake Girardot recently opened an issue to flag any changes to
man_made=survey_pointfeatures. This is now a compare function in OSM-Compare ready to flag the first such change in OSMCha.
- If you have any other niche use case of OSM-Compare that you would like to use it for, awesome!! Please feel free make an issue and I can help you write a compare function in OSM-Compare that you can use as a reason in OSMCha.
Join us in validation
If you are new to OSMCha and not sure where to start, going through
review_requested changesets in your local area is great way to begin. This is a new feature that landed in iD editor on OpenStreetMap. These are the steps I would follow:
- Set the OSMCha filter for changesets with the reason
Review requested. This link will set it for you.
- Open the filters and modify the
bboxto your local map area that you want to review.
Applyto view the list of changesets that was requested for review
- Go through the changesets and submit feedback to the mapper by commenting on the changeset on OpenStreetMap based on your assessment.
During recent months, we have seen a lot of new user activity and pokemon targeted feature mapping on OSM. This correlated with observed/assumed association of pokemon spawn points with OSM features by PokemonGo players.
From my observations on reviewing changesets, some features are mapped very specifically for specific pokemons. Some of these features are parks, meadows, waterbodies, gardens, protected area’s, village_green’s, natural reserves, golf courses, cemeteries and forests .
So I made these comparators in OSMCha that flag a new user’s(<=10 changesets) changeset if one of these features have been added brand new(Version 1).
- This reason flags a changeset if there is a creation of these features in it
- recreation_ground, village_green, park, nature_reserve, protected_area, national_park, garden
- This reason flags a changeset if there is a creation of these features in it
- natural=water, water=, landuse = pond or reservoir, waterway=
- This reason flags a changeset if there is a creation of these features in it
- recreation_ground, park, pitch, playground, golf_course, meadow, grass, cemetery, grass, forest
One can combine reasons in OSMCha and look at changesets now, add bboxes to specific areas and share permalinks. Ex: All pokemon based reasons in Germany.
From our experience validating, a lot of this mapping is really good and rarely do we find any suspicious edits. Let us know if you find these useful or if I need to tweak them based on changing Pokemon trend.
Screenshots of some really interesting mapping activity I have come across
It is always interesting to review the latest map edits in your local area. It can be fun to send a welcome message to a new contributor or track changes to a neighbourhood that a fellow mapper has surveyed. It's an opportunity to both learn from an experienced contributor as well as teach someone new a helpful tip to make mapping more engaging. In a more rare case, these tools can help investigate some missing data or suspicous mapping activity.
Here's how you can start using OSMCha in your area:
Step 1 : Understanding and using Filters in OSMCha
Currently available filters in OSMCha
Introduction to filters
- These date fields are based on changeset closed time as on OSM. We can use date range to narrow down changesets (yesterday, last week)
Creations, Modifications and Deletions
- This is a simple count of type of edits in a changeset
A changeset is suspicious when the changeset is flagged by one of the reasons seen below. These are compare functions that flag a certain type of edits. I will go into more details about how compare functions work in OSMCha and how we can use them to flag specific edits. This has to be an another diary post.
White-list - When you login into OSMCha using your OSM credentials, OSMCha creates a very basic profile that consists a list of changesets from OSM users you wish to not see in your search. This is a personal custom list for each OSMCha user/reviewer. You can add a user to the white-list when you are in that user's changeset.
BBOX - This filter allows to easily give an area of interest we would like to validate. In this case, you can zoom into your particular neighborhood and OSMCha retrieves only changesets whose bbox falls on the area you have given.
These filters give a lot of freedom and flexibility to narrow specific type of changesets we would like to retrieve from a specific area. In OSM one can already use the
History tab to see the changesets that overlap with the area on the map but OSMCha adds a lot of filters to this idea to assist the reviewer.
Step 2: Validating your neighborhood
Using bbox filter to select area of interest
When we click the search button, OSMCha presents us with a list of changesets that it thinks fit into our search criteria. An embed version of changeset-map shows geometric and feature tags of edits on the map.
Along with the contextual information where the edits are on the map, geometric edits to the feature, you can also click on the features in the changeset map touched by that changeset to see previous version feature tags and current version feature tags.
Did you try using OSMCha? What worked, and what did you not like? Let us know in the comments below.
If you come across a bug you would like to give feedback on, use OSMCha-django repository to file an issue. An issue with overlap of global changesets when using bbox filter in OSMCha has been already raised in my previous diary post. We are working on to best resolve this issue.
If you would like improve detection of a particular type of feature edit, check out our OSM-compare repository and open an issue.
OSMCha is a an open source changeset exploration tool originally created by Wille Marcel. Early 2016, few of us at Mapbox were interested in using this tool for trying out validation on a changeset level. Over the course of 2016, we made several improvements to the tool. As of this morning we reviewed more than 23000 changesets and found 1150 to be harmful to the map. OSMCha database consists useful changeset metadata such as changeset ID, username, editor used, changeset comment, source, imagery used, and timestamp.
You can download a CSV of all the reviewed changesets here. For community members who are interested in validating the map using OSMCha, our validation guide can be a good starting point in understanding the tool, how we use it and validate their own neighborhood.
Few things to note
OSMCha does not parse all changesets from OSM. There are a few that go unparsed each day because of various edge cases that we are working on fixing. So do not take numbers on OSMCha as absolute but as near accurate estimates.
Some of the mapping activity marked as harmful in OSMCha are not necessarily harmful. Undiscussed, unannounced imports in OSM are constantly tracked and reverted by the DWG. These edits to the map do not necessarily have mapping mistakes in them but were found to be uninvited into the map to maintain a data import protocol, accuracy on the map and local community accord.
Hence, mass deletion of above imports in revert changesets by DWG cleanup accounts like Woodpeck_repair are also marked as good edits. These can be ignored by filtering out repair accounts.
The reviewed changesets were from random places on the map and are not specific to any place. For area specific filtering we can take advantage of bbox filter in OSMCha or filter manually as the CSV contains the bbox information for each changeset.
Since we have a big dataset of reviewed changesets, we can find correlation between harmful changesets to find patterns of vandalism on OSM. I did a basic analysis using a recently added metadata filter in OSMCha stats page with which I have come to below estimates.
Editor wise breakdown of changesets marked to be harmful
Editor wise breakdown of changesets reviewed
Filters we found to be successful
These are percentage of harmful edits observed against the number of reviewed.
iD+suspect word : 14.1%
iD+mass deletions : 7.9%
potlatch+mass deletion : 5.8%
JOSM+suspect word : 5.8%
JOSM+mass deletion : 4.9%
Maps.me : 3.7%
- Suspect word filter flags changesets with apple, google, nokia, here, waze, tomtom, import, wikimapia as words in changeset comment or source.
Having a database of OSM edits that are classified into good and harmful can help future endeavours into implementing smart anamoly detection tools and machine learning algorithms to better protect the map.
We are looking forward to continue validation using OSMCha, refine OSMCha changeset flagging heuristics, collaborate with the community with better open tools to protect the map.
Let us know your thoughts, how this can be taken forward and share with us your insights to improve feature level detection.
As a follow up to our previous reviews of Maps.me user edits, these are the Maps.me issues we have observed during the month of November during validation using OSMCha. Documenting the errors observed can initiate a conversation on improving user experience, information to the users, characteristics of mapping activity from users. This in turn improves OSM.
- Reviewed: 1078 CSV
- Suspicious changesets: 46
- No issues: 1032
We commented on these changesets:
my homein the name tag: 1
- Added 76 fuel_stations: 1
- Added ton of outdoor shops: 1
- Added attractions for address and buildings: 1
- Added amenity=school for bus stops, canteen, parking lot due to the presence of an adjacent school building: 1
- Added Korean names in the
- Added attraction tag for addresses: 1
- Added description in the name tag: 1
- Added ton of sports shops and other POIs: 1 - Cleanup
- Added 101 benches: 1
- Added 133 confectionery shops: 1
Community members commented on these changesets:
- Wrong naming to the tags: 1
- Irrelavent information: 1
- Added personal name to the supermarket: 1
religiontag but in
name:rutag pointed that this is Ukrainian church: 1
- Added numbers to the name tag: 1
- Added many view points: 1, 2
- Added weird names for post offices like 'Timber', 'Hostel', 'Grocery': 1
- Added convenience shops at the same place: 1
virusin the name tag of all the neighborhoods in Vietnam: 1
- Added number of bus stops together: 1
- Duplicate POI's, community reverted the changeset: 1, 2, 3
- Added number of view points with same name: 1
- Added numbers to the name tag: 1
name=*tag edits with personal naming 1 2, incorect language input, description, alphanumeric names 1 2, single letter names, weird names, virus in name tag, numbers.
- Dense and implausible number of POIs being mapped like view points, attractions, benches, convenience shops etc.
- We have not observed partial uploads and changesets without any changeset comment are very few.
- Users are not responding to changeset comments.
It would be great if the community comes together, share thoughts and ideas to help the developers fix these issues in future. Issues can be filed in the Maps.me repository.
During the month of August this is the approximate changeset percentage breakdown between editors that uploaded user edits into OSM:
As a follow up to our review of sample edits from Maps.Me, we decided to randomly review changesets during August to understand the difference in quality and contributions from this editor from our previous review.
Changesets reviewed: 475
Problematic changesets: 12
Minor issues: 33 Raw notes
No obvious issues: 430
The edits looked like this on the map
2.Changeset with 91
toursim=attraction POIs. A community member commented and DWG reverted the changeset.
Density of artworks as seen on OSMCha
3.Changeset with 78 ATMs. Community reverted this changeset.
4.Changeset with a lot of of
tourism=camp_site tags in one area and some are on water. We commented on this changeset asking for clarification.
5.Changeset with a lot of
tourism=attraction tags in names such as
gym. We commented and reverted this changeset.
6.Changeset with POIs on roads. We commented on the changeset to let the user know of this.
7.Changeset advertising apartment in demand. We commented on the changeset.
8.Changeset added a bookshop and an outdoor shop at a stadium. We commented on the changeset asking for clarification if these are temporary stalls.
tourism=camp_site POIs to map railway bridges, switches and crossings. We reverted the changeset with a changeset comment.
amenity=bus_stations over a residential area. A community member commented and reverted the changeset.
Changes observed since our previous review
We have not observed
name=*modifications from Maps.me users.
We have not come across partial upaloads and changesets with no changeset comments were very few.
But we have continued to observe below issues:
Similar to our first review, there were users using tourism tags instead of the appropriate tags.
Users are not responding to changeset comments.
Comparatively, we saw less number of problematic changesets with respect to the number of changesets we reviewed. Let us know what you think of the editor and where it should improve.
Number of new contributors on OSM Source: https://wiki.openstreetmap.org/wiki/Stats
There has been a significant spike in new contributors to OSM that coincide with the launch of edit feature in maps.me. Do you think it correlates with new user contributions using Maps.me? Let us know what you think.
Continuing from our previous weekly round-up, we would like to keep you posted on some of our observations this past 2 weeks.
We commented on the following changesets:
Deleted a tertiary road: Changeset.
Deleted buildings: Changeset.
Added duplicate buildings: Changeset.
Building, parking lot and service roads deleted: Changeset.
Some of the POIs added are on roads and river: Changeset.
Created duplicates of highways and buildings: Changeset.
highway=tracksand waterways: Changeset.
Community members commented on the following changesets:
Added a POI in the middle of the sea: Changeset.
Changeset using Google as source: Changeset. A DWG member commented on the changeset.
Added bus station over residential area: Changeset. A community member commented and reverted this changeset.
operatortag with personal name for bus stops: Changeset. A community member commented and reverted the changeset.
Look forward to these posts as we wish to continue posting weekly round-ups to inform the community on our findings on OSM.
We're going to be at SOTM in Brussels next week. Catch up with @planemad, @jinalfoflia, @ramyaragupathy, @pratikyadav and @geohacker on the latest data team projects at Mapbox!
Following up on a recent series of posts that summarized the suspicious changes on the map, an interesting question that we bring to the community - what is a good amount of time to wait for a response from an inexperienced mapper before fixing or reverting their edit?
You can see examples of such changesets that do not seem right on these posts:
- https://www.openstreetmap.org/user/manoharuss/diary/39050 (Maps.me review)
Observations from last week
- User added unexplained nodes with alphanumeric names: 1, 2. A community member commented on these changesets and reverted them.
The community is great at responding to mapping issues but there are few issues lying under the rug. When we see unexplained edits without a clear comment or source, we do not go and fix them immediately, but let the user know the best practice and try and encourage them to to become a better mapper.
During the last 3 months the data team at Mapbox has commented on around 78 changesets and fixed only around 10 changesets. Over the course of last 1 year we have commented on around 250 changesets that did not get a reply from the user (quick map of these changesets).
It is a very subjective question to ask how long do we wait before we take the plunge into fixing or reverting changesets like these which may be suspicious, but is difficult to verify from the original mapper. Happy to hear how you approach this issue.
Sharing this open dataset of Canada road networks from the Canada Open data website. The wiki mentions Statistics Canada(StatCan) as one of the data sources, but the StatCAN wiki page, provides a link to an outdated portal where road data is available only until 2010.
- Provider: Statistics Canada
- Published date: 2015-05-27
- File formats: Shapefile, GML, JSON
- Frequency: Mentioned "As Needed"
- Contact: http://open.canada.ca/en/forms/contact-us
- URL: Canada road networks files 2015
- License: http://open.canada.ca/en/open-government-licence-canada
The road network supplied includes road classes, road names, city etc. Through a preliminary look in QGIS I have these findings.
Coverage as seen on QGIS, road classification on the left
The road classification is represented in numbers in the
Major highways are given under class 12 in the attribute table
There were few concerns on OSM wiki page of StatCan regarding geometry of the roads. But the wiki page seem to be outdated and the concerns were in regards to the older data (2010). The geometry of the roads in this data set seem to align with the OSM layer in QGIS.
Roads in Toronto overlayed on OSM layer in QGIS
The street names on the data seem to be same as on OSM as well
Coverage in Toronto without OSM background
This seems like a great dataset to evaluate the quality of data and coverage of roads in Canada. Would be great to hear from the Canadian OSM community of any previous work using this dataset that can be taken forward.
Turn restrictions are the last missing piece in the puzzle to make OpenStreetMap ready for accurate routing. With Mapillary street photographs and their traffic sign recognition, it is easier than ever to start mapping missing restrictions onto the map.
At the data team in Mapbox, we have been experimenting with creating mapping tools to simplify such efforts and after adding over 1,200 turn restrictions in 30 US cities, we are ready to start mapping in Canada with the help of the local mapping community! Our team will focus on the following 5 cities: Ottawa, Toronto, Montréal, Vancouver and Calgary
How to contribute
If its your first time mapping turn restrictions, read the guide to mapping turn restrictions using Mapillary to understand different scenarios and special cases.
Use the OSM navigation map to compare traffic signages from Mapillary and the map data for potential restrictions to add. You can also review the photograph and mark the restriction as valid or invalid.
Marking a detected no left into a oneway as redundant restriction on the OSM navigation map
More Mapillary = More detected signages that can be mapped. Check the Mapillary coverage in your city and fill in the gaps! Calgary currently has the poorest coverage amongst the 5 cities. This tutorial will help you can get started with Mapillary.
We estimate it to take 1 full week to review all the detected turn restrictions in the 5 selected cities with just our team. But could finish it off sooner and add more restrictions with a wider participation. These are the current number of restrictions present that were queried using Overpass:
The road ahead!
It would be amazing to have the Canadian Mapping Community to help us out in making the map of Canada more navigable and enhanced. We would love to hear your ideas and thoughts on how to make our existing workflow better. Interested folks can contribute to the Mapillary coverage in Canada which would certainly help us add quality data onto the map! Let's all join hands in making OpenStreetMap the best!
TLDR: Most contributions from MAPS.ME editor are constructive and has significantly increased new contributors to the map. With any new tool, there are common mistakes that is made, many of which could be easily prevented by simple improvements to the editor and awareness for the users.
Following recent reports of bad edits, we decided to review a random sample of changesets in detail last week to understand the quality of contributions from this new editor.
- Changesets reviewed: 215
- Problematic changesets: 8
- Minor issues: 19 (Raw notes)
- No obvious issues: 188
Looking at these changesets in detail can give some insights on what improvements could be made to help new mappers contribute constructively to the map:
- User added a castle that do not exist which was later deleted by another user.
- User added a lot "artworks". A community member commented on the changeset but the user did not reply yet.
- User added 211 hardware shops. I have commented on the changeset but did not get a reply.
- User added a duplicate POI. A community member commented but did not get a reply.
- User added 29
tourism=viewpointtags to all kinds of things. I commented on the changeset.
Department stores everywhere looked like this. I commented on this changeset.
tourism=viewpointtags to 23 nodes in one area. I commented on this.
- The problematic changesets seem to indicate that new users may be unaware of best practices of mapping, or do not know their edits are seen by everyone else
- Users do not respond to changeset comments
- Users tend to use an incorrect tag like
attractionif they don't find the appropriate category for their POI
- Changing local names to non local names was observed consistently
- Partial uploads / lack of changeset comments
- Positional errors
- User mistook adding POI for a personal bookmark
For monitoring and reverting changes from MAPS.ME contributors, Ilya Zverev has made a convenient web tool http://mmwatch.osmz.ru/ , and a reverting tool http://revert.osmz.ru/ to track and fix edits by country. In the monitoring tool, one can see the latest edits and tags edited, and also mass revert multiple changesets. The revert tool allows easy revertion of a list of changeset ids.
What are your thoughts on the editor?
Here is this week's collection of suspicious mapping observed between 20 - 24th June.
- This changeset from a new user was observed to be deleting POI's, beach, public toilets and adding a random square without reference to the imagery or any mentioned source. This was reverted.
- This changeset changed name of the Paracel islands from Chinese to english, changed the locality from Chinese to Vietnamese, added a traingle on the map with no reference. This was reverted by the DWG.
- This changeset was observed to be deleting a lot of buildings and roads.
- This changeset deleted a lot of service roads.
- In this changeset the nodes had
- This changeset had given
building=yestag to resendential areas instead of the individual buildings.
- This changeset deleted buildings using iD editor with no explanation in the comment.
- This changeset deleted private roads using the iD editor. This was reverted by the community.
- This changeset added address tags to every single object in the changeset including the nodes. The cleanup of this changeset was detailed in this diary post.
And ofcourse, look forward to this roundup again next week. If you observe any suspicious mapping, comment on the changeset and let the mapper know. Mistakes happen all the time.
Weekly observations between 27th to 30th June can be found here.
The amount of data on OSM sometimes boggles the mind, there was 23471 new changesets just yesterday. With so much great mapping work being done by the community, there's always a possibility of a few bad apples. So, here's a collection of a few that I have stumbled across last week (13 - 17 June).
This changeset in which a number of
highway=footway & footway=crossing & crossing=zebratags were deleted and nodes on highways with
highway=crossingtags were added in downtown Atlanta. It is to be remembered that both these methods are suggested in the OSM wiki for mapping crosswalks.
This changeset deleted roads in Ethiopia using iD editor. Also 5 harmful changesets from 2 other users from whom one was deleting buildings and the other was editing features without any reference imagery or mentioned source.
This changeset gave non-latin names in
nametag in London and this was reverted.
In this changeset, nodes had attributes meant for the ways.
This changeset fixes a user changing
Stumbled on anything suspicous on the map this week? Do drop the mapper a message, for all you know, it was just an honest mistake :) Look forward to another roundup next week.
I have been mapping my home town of Gudivada. This is where I grew up and did all my schooling. In the Telugu language, "Gudi" means temple, and "vada" means a settlement or a town - Gudivada has many Hindu temples.
When I decided to map, I expected that since this is a small town there would just be a node with a name, and not much else. I was wrong though - trunk, primary, secondary, and tertiary roads along with many points of interests and buildings around the center had been well mapped, thanks to the active OpenStreetMap community.
I focused my efforts on mapping residential streets, and adding places like hospitals, temples and schools I was certain about.
Before and after
Here are the residential roads I added within Gudivada. A number of features such as water bodies, public water tanks, school names and area names which I have added most recently are not visible in the 'after' gif due to OpenStreetMap not rendering those objects at that zoom level.
Closeup of my edits
Here is a closeup of my edits in my hometown. One issue I encountered was not being 100% sure about road names so refrained from adding local names to the roads. For example, as I was mapping, I came across "Eluru Road" which goes across the town - my hunch is that this road has a different official name.
What I added and modified:
- Missing residential roads were added
- Re-classified many roads to Tertiary roads based on road classification documentation
- Added places like schools, school grounds, property boundaries, water bodies, places of worship, stadium and other amenities.
- Misaligned roads
- Added amenities for existing buildings
I've enjoyed mapping my hometown and added numerous features which I am intimately aware of. My plan is to add more details based on satellite imagery and local knowledge, and do a field survey the next time I visit to ground-truth Gudivada data on OpenStreetMap.
I'd also welcome any comments others in the community have to improve what I've mapped, or tips for mapping.
Locality mapping and updating addresses
Hello everyone. I am a mapping enthusiast and I seek to add geocoded data for Sri Ramachandra nagar in the city of Vijayawada. It is a city with about 850,000 population and it is a major city in the state of Andhra Pradesh, India. I am familiar with the area around my home.I am looking to tag places of importance in the city, create and edit data in Sri Ramachandra nagar. I have started by tagging full addresses for major establishments I know of. I would like it if anybody in the area helps me out with the door numbers and details of newer businesses and residential buildings.
I have started by tagging around Ayush Hospital. For anybody who is interested to map in Vijayawada, I am willing to collaborate. Usually ton of educational institutions, hospitals have websites. In the business website you can find the full address with door numbers and pincodes. That should help you to enter the information into OSM. These businesses have constant addresses and should help in tagging addresses. Cheers.