Mundi app changesets revert

Hello Israel Community!
Around a year+ ago some Technion students released a new OSM editing application called Mundi. The idea is similar to the well-known StreetComplete app.
At that period several new users started actively contributing with this app and some even produced 100+ commits.
The problem is that the quality of these commits was usually very bad and, IMHO, contribution was minor compared to the garbage they produced and somebody would need to cleanup/fix.

I contacted the app developers (mundi.mapping@gmail.com, Yosef Break on LinkedIn)
and asked them to fix issues and switch to test osm servers until app is stable and usable.

Few examples of bad commits:
148706974 adds weird footway=bus_stop tag (and there are more changesets like this)
148706952 replaced totally valid highway=crossing with highway=footway
148725685 added kind of correct but disconnected from any ways node
146461806 added not existing “num_of_lanes” tag
118009450 added barrier in the middle of nowhere

What I’m asking is community approval to bulk revert of such commits since editing all the mistakes one-by-one is way too much time consuming.
Alternatively, I can try to run overpass queries for known issues and try to fix them in bulk, but still issues like disconnected nodes will be hard to fix.

I’d start with 3 months old commits made by (reverted) דלית | OpenStreetMap

More “Mundi users” to revert (all of them added total nonsense): list

1 Like

I have tried contacting the developers due to mapping of a big amount of gates in a settlement that popup up on the IHM map that were not connected to any way. I got a response about half a year layer with no accountability or any desire to improve.
They said they are using OSM like any other app.
I agree about the garbage and the usefulness of the data and I’m in favor or removing it.
The last mail I sent was January this year and I didn’t got any response to it.

Here’s the content of the mail thread, my question are in hebrew:

  1. האם זה נעשה ע"י האפליקציה שלכם

Anyone can enter anything they wish into OSM, and there’s no way for us to track or be aware of it. The app doesn’t trace users’ mapping values or hold them responsible for it.

Therefore, blaming the app for inserting junk into OSM is untrue, and all app users have to input the data manually. The data which we collect includes the total actions and types performed by a user for global statistics.

If you suspect any misuse of the app by a users or bot, please inform me, and I will investigate the matter for you.

  1. האם האפליקציה מוסיפה מידע ל-OSM

Yes. It’s not confidential; you can discover it in the app description.

  1. מה הבקרה שאתם עושים על המידע שהאפליקציה מכניסה?

I had the same question during the initial meeting, and here’s the response: “It’s the same as OSM = Anyone can enter anything they wish into OSM.” The app is simply another entertaining way to map objects around.

Each user has to use their personal OSM account to map things. Furthermore, users are typically junior high school students who passed a mapping guide with a Technion instructor.

.In case you want to make an appointment with the relevant person, let me know. I’ll inform them

.Be aware it’s not completely a project yet, and there’s always a window to improve

Moreover, it’s important to note that my response might not be entirely accurate and could potentially include mistakes. I am a student working on the IOS version.

Thank you for understanding,

Yosef.

Which I replied and got not answer:

The reason I asked the original question is that a lot of gates popped up on our map which uses OSM (this was a long time ago and it took me a while to remember why I sent the original email).

Below is a link and image to how it looks - a lot of these gates are not part of any road, which means that their mapping is probably incorrect or partial.

If this app is given to high school students I would like to ask that the instructor will do some verification of the work the students are doing, at least to verify that their first edits are valid and valuable.

While we too use OSM the same way, there’s a higher barrier to adding points to the map (beside login, the user needs to do a few clicks to upload a point to OSM), and especially for gates we try and find the closest highway in order to place the gate on it.

If the purpose of the app is not to add valuable information to OSM, I would recommend using their dev server and not the production server.

image.png

1 Like

Hi! DWG is on this one. Ticket#2024062710000027 if anyone wants to email anything outside of the topic. I’ll monitor here too.

Elliott
OSMF Data Working Group

1 Like

Found out Mundi could add empty keys - here name:en and name:he both have empty value:

I found it while trying to create an Overpass query to find empty values… I got only this, and I have no idea how I did it because I can’t recreate it: new Overpass queries always return nothing even when I try using exactly the same criteria I remember using.

Or in short: I don’t know how to find more.

I have some code that parses an OSM file, do you need me to find elements with empty values in Israel?

Not particularly, I just pointed it out because it’s another thing Mundi messed up. I’m probably not going to go around fixing all occurrences of this (focusing on other things).

Edit: can you explain though, what do you mean by OSM file? The planet file?

Geofabrik has a daily extract of Israel which can be used to query stuff when loaded into memory by code.
https://download.geofabrik.de/asia/israel-and-palestine.html

Hi, I just noticed there are a lot of crap nodes created by this user with Mundi: Changesets by יהב מושקה | OpenStreetMap

Do you have a quick and easy way to delete all these nodes? They’re mostly bollards. Deleting them all manually seems too tedious and error-prone, so I figured the DWG might already have tools to take care of it.

1 Like

@NeatNit ,
I just revert all the changesets for specific user. they are usually 110% junk. I can do this today.

@ElliottPlack , do you have any semi-automatic way to find all users by commit comments? (and optionally also block them, in case they’ll decide to use this app again)

I am investigating this, thanks for the reminder. It appears that all the changesets have the created_by=Mundi App tag, so that is helpful.

There is a tool I can use to mass revert all or some of a user’s changesets, but we have to know the user’s name. What I am unsure about is a method to query all changesets with a given comment or created_by tag. Some of the ones you’ve listed have 100% Mundi tags. If you can help me compile a list of those users that have no good edits whatsover, those are very easy to do. However, if there are users that have a mix, we’ll need to figure out how to subdivide the good from the bad.

Actually, I’ve started a list that you (or anyone here) can edit.

IMHO, you can safely revert them. The overal percent of good contributions is really negligible compared to garbage.

actually finding these users is the most problematic part. We randomly discover them from time to time.
It might be an overkill, but we can try to query all changesets in Israel in specific period of time, looks like most of changes were done in a pretty short period of time. Looks like they had a “mapping party”.
You can take 2022-03-10 as an example.

@ElliottPlack , I just did few queries by known tags used by MundiApp and, sadly, discovered around 40 not-yet-reverted users… I believe there are more.

Will appreciate if you can at least revert all of them to lower the “density of nonsense” on the map, and then we’ll try to catch the rest.
(I can do reverts myself, but regular users have limits, so process will be very long)

Please “welcome”, new mundi generation - open_street_mundo

Please help monitoring this

Subject: Feedback and Support for the Mundi App and its Contribution to OSM

Dear OSM Community,

I hope this message finds you well.

Recently, I noticed concerns regarding the authenticity of edits made using the Mundi App, with some stating that many of these edits are problematic. I would like to provide some background information about the app and its purpose:

  • The Mundi App was developed in The Technion as part of a citizen science project in schools, aimed at raising awareness of navigational accessibility issues in the built environment for blind pedestrians. You can read more about this research project here: https://www.researchgate.net/profile/Pengyuan-Liu-3/publication/362752226_Proceedings_of_the_Academic_Track_at_State_of_the_Map_2022/links/62fd80d5ceb9764f7204553c/Proceedings-of-the-Academic-Track-at-State-of-the-Map-2022.pdf#page=19.
  • As part of this initiative, students explored their neighborhoods to map valuable accessibility-related features that are important for the visually impaired community and are often missing in OSM. During one particular campaign, several hundred students mapped simultaneously, which unfortunately resulted in data duplication. This was an isolated incident, and we have no plans to repeat such a large-scale campaign.
  • Typically, edits made through the app are localized and verified by a supervisor to ensure data quality.
  • We believe the app has significant potential to benefit the pedestrian community broadly and the visually impaired community specifically. Once the app reaches a stable version, it could also serve as a valuable tool for the broader OSM mapper community.
  • Given the increasing reliance on OSM data by navigation systems—some of which are tailored for disabled communities—the Mundi App contributes to the overall enhancement of such services. You can read more about the importance of this type of data: https://journals.sagepub.com/doi/full/10.1177/23998083241256402.

We value your thoughts and would greatly appreciate your feedback. Any suggestions or comments you might have on improving the app and avoiding similar issues in the future would be invaluable to us.

Thank you in advance for your time and support.

Best regards,
Dr. Sagi Dalyot.

The response time of the app developers and owners is far beyond a reasonable timeframe.
The lake lack of responsibly and ownership of the issues the app creates to the community is unsatisfiable.
I would recommend reading this thread thoroughly and coming up with ways to address the concerns raised here.
Otherwise, the only possible direction the community can go is to continue treating this app as a hobby project and not a valuable asset and continue revet edits made by it.
If this wasn’t clear enough from the thread comments, I hope it’s clear now.

2 Likes

Hi @open_street_mundo (Dr. Sagi Dalyot).
It’s truly inspiring to see your work and field of research.
I invite you to chat with the community on our Telegram channel: Telegram: Contact @OSM_Israel so we can work this out together.

BTW, this is a website explaining all the different aspects of this app and research:

Thank you :slight_smile:

@YaronS, thank you for your assistance.
I truly believe that the app has a strong socio-urban contribution. I would be happy to talk to see how we can solve this issue (unfortunately, I do not use Telegram).

1 Like