While commercial map providers like Google, provide map data in local Indian languages, the local language labels are usually transliterated from English, resulting in errors. OSM, Wikidata and Wikipedia platforms provide a way to improve the local language maps leveraging multilingual labels through Wikidata identifiers on OSM, through semi automated updates. For good quality maps on Wikipedia and OSM, the Wikidata location needs to be accurate and Wikidata identifier should be uniquely mapped to the actual OSM geometry. I worked on removing the mismatches for villages of a district in Andhra Pradesh. I have documented my experience as clearly as possible, so that even users with less programming and tools skills can contribute to the work. As an example, identifying and fixing errors in Wikidata location is given below.

Identifying and fixing error in locations on Wikidata pages

The geodata is presented in Wikidata page and corresponding English Wikipedia article page using OSM as background map. If one notices that the marker is not near to the names identified on OSM map, then there is possibility of an error. Even if the name is identified on OSM background map, selecting different zoom levels allows checking whether the place is in the correct location.

The correct geolocation is obtained from Open Data compliant StateGIS portal run by Bharatmaps.

After Wikidata coordinate location is updated with the correct data, Wikidata page and the corresponding English wiki article show the correct map.

Full article covers how to identify more complex errors based on the geographical distance between Wikidata location and OSM node location and also when the Wikidata and OSM geometry that is associated are outside the area of interest. Sample Wikidata Sparql scripts leveraging Sophox interface to OSM, which can be easily edited are provided to help contributors with less programming and tool skills to contribute to these efforts.

For more information, check out Improving geodata accuracy on OSM and Wikidata (Full article on OSM wiki) and let me know your feedback.

I did lot of contribution data analysis on Wikipedia. I wanted to do similar thing for OSM. Finally, here is my first analysis. I downloaded the fuel outlets data as of 2023-07-07 from OSM in Andhra Pradesh using overpass-turbo query. I used google sheets to analyse the data. Here are some results.

Summary of geometries

Total : 504

• node:373 (74%)
• way:131 (26%)

Year of last touch

Year    Count
2008    1
2011    8
2012    17
2013    23
2014    6
2015    23
2016    29
2017    23
2018    59
2019    52
2020    48
2021    54
2022    83
2023    78

Unique contributors

year	Unique contributors
2008	1
2011	5
2012	5
2013	6
2014	2
2015	6
2016	10
2017	10
2018	18
2019	19
2020	28
2021	8
2022	24
2023	10

Top 10 contributors

Top 10 contributors (as per last touch) accounted for 54.76% of contributions. There were 132 contibutors in all.

Total    User
57    arjunaraoc
53    Srihari Thalla
53    uknown-gryphus
22    vaibhavnz
19    Chaitanya Kumar CH
18    gunmac
15    IbnTēšfīn
15    sk_trav
13    Heinz_V
11    Kesari

Brand info

Count    Brand
328    Unnamed
 76    Indian Oil
 28    Hindustan Petroleum
 25    HP
 19    Bharat Petroleum
  9    BP
  4    Indian Oil Petrol Pump
  3    IOC
  3    Reliance
  2    Bharath Petroleum
  1    Bharath
  1    Essar
  1    IOL
  1    OIL!
  1    Petrol Ofisi
  1    Petronas
  1    Relience

There were 4168 operational petrol pumps in Andhra Pradesh as of 1 Jan 2022. Therefore coverage on OSM is about 12%. So there is lot to add and cleanup.

I added 52 fuel filling stations which is 11.5% increase over earlier count of 451 across Andhra Pradesh. I spent about 10 hours logging the points of interest (POI) and another 10 hours updating them. Here is my backstory or how-to-do guide. I have been an OSM mapper from 2015, having learnt basics from PlaneMad and received mentoring from Heinz Vieth. I have been mapping occasionally about my neighborhood and places that I visit. I also installed Osmand on my smartphone and explored its navigation and trip and POI recording features. In my recent trip to my native state, I tried to record road side fuel filling stations as I drove on the roads as a passenger in car or bus. I used josm to update the OSM.

POI recording

Setup Osmand and update maps to their latest versions. Install “Trip recording”, “Audio/video notes” and “OSM editing” plugins. Setup Trip recording profile to “Driving” or “Public transport” as the case may be. In the “OSM editing” plugin setting enable “Offline editing”. Setup OSMand to navigate to your destination. As you pass by a fuel filling station, record a POI by tapping with your finger, if the feature is not already visible on the map. Label it with a short label such as “petrol” or “hp petrol” (for filling station with brand Hindustan Petroleum for example). Continue this exercise for all points of interest. Even if you type the same name next time, the app will append the serial number to that name while saving it.

Create OSM file containing smaller set of POIs

Transfer the stored POIs from “My Places” menu of Osmand using the share button to your email account. Setup JOSM on your computer. Install “todo” plugin. Open the exported POIs file received via email (example: favourites.gpx). It will show two layers “favourites.gpx” and “Markers from favourites.gpx”. Select “Convert to data layer” option for the latter and you will end up with “Converted from: Markers from favourites.gpx” as your new layer, with the older “Markers from favourites.gpx” layer gone. Remove checkbox from “Discourage upload” corresponding to this layer. Save this layer as “favourites.osm” file. Depending on your trip, you may have lots of POIs to cleanup. I ended up with about 130 POIs for about 8 hours of my trip, which included other facilities as well. To make it easier to work and as “todo” plugin does not have the ability to save in progress status, create new files with each one containing approximately 10 POIs (by using copy from the original file and saving them in a new layer with “paste at source position”. After the new layer is saved, delete the POIs already copied and proceed to iterate for a new set of 10 POIs. You can name these files as ‘favourites part 1.osm”, “favourites part 2.osm” etc.

Cleanup with satellite imagery and upload

Start JOSM. Load the “.osm” file containing POIs to be cleaned. Select Edit -> search and enable “select all objects”. After all objects are selected, add them to “todo” plugin with “Clear and add” option under the Add menu button. Add Satellite imagery layer from one of the satellite image options like Bing.

Zoom to the first POI on your “todo” list. Adjust the zoom so that you can see the nearby highways on the satellite image. Select download in current view to see the already existing features of OSM. Locate the filling station from the satellite imagery, with its usually distinctive features of service roads from the highway and a rectangular reflective frame in the centre surrounded by building like features. If POI already exists, delete the recorded POI. If it is not present, move the recorded POI to the location of the filling station as per the satellite image. Apply preset of fuel from “Transport Group/Car” icon. Update the name as “Filling station” and update brand name if available from the label. Mark the item in the “todo” list. Your view will shift to the next item as the center. Repeat the steps.

If you do not find evidence of filling station from the satellite image, because the image is old and the construction is new, delete the POI from the layer, mark the POI and proceed to the remaining.

Once the cleanup is complete, upload your edits to OSM and use remarks such as “Update of facilities near highway” as “Description” or “brief comment” and specify the data source as “Drive-by survey during MMM YYYY and ‘Satellite name’ imagery”.

Repeat the above for all the “part” files and you are done. You have used your travel time productively and also contributed to a better OSM.

Notes: 52 new filling stations Interactive Overpass-turbo OSM map (patiently wait for Overpass-turbo to process the query)