Too easy to create notes without setting a location

[angoca]

Not sure if the origin is from this page or any fork, but I see many notes in the middle of the Atlantic Ocean for Iranian places.

Some note's id I have closed:

2735935
2723484
3451396
2057596
1762208
1762208
3443951
1840542
3408709

[ArcaneDots]

Thanks for catching the invalid entries, I had a feeling that might be a problem with all of those duplicates.

…

On Sun, Dec 11, 2022, 12:09 AM Andres Gomez Casanova < ***@***.***> wrote: Not sure if the origin is from this page or any fork, but I see many notes in the middle of the Atlantic Ocean for Iranian places. Some note's id I have closed: 2735935 2723484 3451396 2057596 1762208 1762208 3443951 1840542 3408709 — Reply to this email directly, view it on GitHub <#103>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAIGQRFK2ZYZSPV7NAPNLBTWMVOYTANCNFSM6AAAAAAS2XG2DM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[iandees]

Thanks for pointing these out. I think these are coming from users of onosm.org moving the pins to random spots on the map. It doesn't look like they're geocoding addresses to these places. It might be a spam attempt?

I'm open to suggestions for how to improve this.

[ArcaneDots]

We should be geocoding the new pin location and then verifying that the address is still valid.

[ivanbranco]

Same is happening in Italy at these coordinates: 41.6900000, 12.7100000

[ArcaneDots]

Please check out my progress on PR #116 which is designed to fix this issue. My solution is to limit marker movement in Leaftlet based on the original's search bounding box. Please read the notes and add your comments.

[ArcaneDots]

Just stumbled over the Iranian version of OnOsm.org. Its source code is hosted on OSMIran on Gitlab. It is a copy of the main fork as of early 2022. The biggest difference is that it supports Farsi which is the main language of the region. Even though Right to Left language should be handle correctly by the browser (via HTML), we will need to verify that everything still works as well.

[iandees]

Just stumbled over the Iranian version of OnOsm.org

Good find!

[angoca]

I created an issue in that fork: https://gitlab.com/OSMIran/note-tool/-/issues/3

[ivanbranco]

Doesn't seem to be onosm.osmiran.ir, their notes have the following text:

onosm.osmiran.ir یادداشتی از یک کسب‌وکار ارسال کرد:‬
نام:
تلفن تماس:
وب‌سایت:
شبکه‌های اجتماعی:
ساعات کاری:
دسته‌بندی:
نشانی:

The ones we found have onosm.org, not onosm.osmiran.ir, and all the fields (Website, Phone number ecc.) in english.

Also seems like it's not even used in Iran, of the last 500 notes, 279 were created from onosm.org and only 5 from onosm.osmiran.ir

[ArcaneDots]

I have create PR #116. It is designed to do some basic validation and geo-fencing of the marker. User can do minor adjustments around the initial point. Anything more is based on the bounding box returned by Nominatin during the original search. That is as long the bbox is larger than the default circle. While outside the circle, any coordinates must fall inside the bbox connected to nearest valid point. This seems to prevent the marker from be left in water or other places for that might be odd for a business to be located.

[ivanbranco]

Now also american notes started to pop-up at the same location:

https://www.openstreetmap.org/note/3584982
https://www.openstreetmap.org/note/3584983

Same is happening in Italy at these coordinates: 41.6900000, 12.7100000

[ArcaneDots]

The underling problem is that the website doesn't currently do any type of validation of the marker moves. My current change tries to limit the distance the marker can be adjust to prevent these huge deviations. I'm still not sure why there are so many duplicate notes getting created. For the moment, I need some feedback of my PR #116 which should prevent part of the problem. I would appreciate code reviews from those who are able and user testing otherwise. I am particularly interested in thoughts relating to creating a reasonable bounding box based on the original user search.

[ArcaneDots]

Are we still having problems with multiple notes?

[ivanbranco]

Are we still having problems with multiple notes?

Less frequently but yes, the latest was 6 hours ago: https://www.openstreetmap.org/note/3658112
In that same coordinates I closed numerous notes.

[Sosha1996]

Hi,
I'm Sosha from OpenStreetMap Iran. I am one of the maintainers of @OSMIran.

First, thank you all for developing onosm.org. This is a very good website for ours and the global community.

We developed a new fork of onosm and translated this to the Persian language. This may also help for other RTL language communities like Arabic or Hebrew.

In this specific case: #103 (comment)
This problem did not arise from our side. But if we can help to resolve, we will do it.

[ivanbranco]

Now also american notes started to pop-up at the same location:

https://www.openstreetmap.org/note/3584982 https://www.openstreetmap.org/note/3584983

Same is happening in Italy at these coordinates: 41.6900000, 12.7100000

I did a test: https://www.openstreetmap.org/note/3845609

onosm creates a note at this coordinates if a user access the website directly to https://www.onosm.org/#details instead of passing from the first step in https://www.onosm.org/

is there a way to avoid this?

[iandees]

onosm creates a note at this coordinates if a user access the website directly to https://www.onosm.org/#details

I'm not able to reproduce that problem. If I visit the site using this link, I am redirected to https://www.onosm.org and asked the enter an address.

[iandees]

Ok, if I go to https://onosm.org/ and then browse to https://www.onosm.org/#details I'm able to skip the location step sometimes. I'm not sure how a normal user of the site would know to do that, but I'm at least able to repro it.

[iandees]

I made some changes in the most recent couple commits (8647919 and 4a04cc8) that should make it harder to fall into this situation. Having said that, the marker's default location in the code was in Italy, so I think these notes are likely coming from another onosm.org fork.

@Sosha1996, can you take a look at the commits I mentioned here and try to pull them into your fork?

[ArcaneDots]

Created PR #125 to prevent user from accidentally creating duplicate notes.

[daganzdaanda]

I've noticed that many of the note clusters (especially in Iran) were at the place node of the city. This makes me think that some users just enter the name of their city on the first tab and then continue to add the details without checking the marker. Seems like the warning message / reminder is not enough in those cases.

Could something be done to force people to check the marker's position if they only gave a city as address?
An idea: Zoom in a bit on the city, but don't allow the user to go to the next step without zooming in more (so that they could realistically find the correct location) and moving the marker at least a little bit.

Actually, I think I may have found a reason for the issue with notes on city nodes: See #127

[daganzdaanda]

It looks like the whole system of setting the location only really works when there is a city, street, and housenumber in the OSM data. If something is missing, people can't move the pin to where the business really is. See also #129 and #127

One big problem is that in most parts of the world, OSM does not have good housenumber coverage. In many parts of the world there are no such things anyway (do Iranians use housenumbers everywhere?) . Also, many places don't have all the street names mapped correctly.

So it may be possible that someone tries to have his shop show up as a note, but can't get it to be at the right position. If they try again and again, they will produce a lot of duplicates.

What could be done to minimize the mis-use of OnOsm?
Should it be possible to move / set the pin much more freely?
A big fat warning in the local language about the shortcomings of the address search?
Maybe remind people to turn on geolocation on a phone, if that would enable them to set a pin close by?

I hope it will be possible to repair this tool -- people in Iran seem to like it, but they should be able to use it successfully.

[claudiush]

Theoretically housenumbers exist in many places in Iran, but commonly addresses are given as description, e.g. "In Tehran, Olympic Village quarter from the crossing at the police station go three streets along Shahriar Boulevard, then right, second house on the left." (I just made this up). So any user adding a note would search using a similar approach starting with the city.

[matkoniecz]

There is some related discussion at https://community.openstreetmap.org/t/osm-notes-around-the-tehran-node/108248/

[iandees]

I made a change in 0b32e6a to add a few more requirements before notes can be submitted. I also adjusted the way map pin movements are required. I don't think I solved all the problems described in this ticket, but I think it's better.