Large amount of personal tags in West Bank / Gaza Strip

Well, it depends. As I mentioned before, some of the previous West Bank/Gaza edits that were of the form “X’s house” were actually edited by more than one MAPS.ME user. In at least one case X was someone associated with the government, so “X’s house” was not unlike a sort of “council office”. When I was cleaning data before I used “more than one MAPS.ME user has edited the data” as an indicator that it probably ought to stay.

That said, some stuff clearly doesn’t belong in OSM (e.g. “my house” http://www.openstreetmap.org/changeset/50106544 ).

Normally I’d also try and contact the users concerned in a language that they’re likely to be familiar with (although the “my house” node obviously suggests they have better command of English than I have of Arabic). Online translation tools such as Google Translate and Bing Translator may not be perfect, but even if it’s as bad as https://en.wikipedia.org/wiki/English_As_She_Is_Spoke at least some words in a recognisable language should draw the reader’s attention. However, with MAPS.ME, there’s always a reasonable chance that they have no idea what OSM is, are just creating personal markers, used a throwaway email account to sign up and therefore never see changeset discussion comment messages. In these cases a 0-hour block message that the user has to read before continuing to edit will effectively stop them contributing (at least from that account)

I can speak Arabic fluently. I tried contacting mappers in both languages. So far I got 1 response, saying the houses are probably just personal bookmarks. Cases like the “council office” example above are exactly why I was reluctant to delete. “X’s house” can have a different meaning locally, or some value.

What do you think the best course of action should be? Should I assume most edits are junk and go ahead and delete anything “X’s house” that was edited only by a single MAPS.ME user?

That’s pretty much what I did last time a DWG ticket came in for them - the exception being where if there was other useful info on the node (like part of an address, for example).

I started the cleanup. With these rough rules:

  • Remove all “my house” type nodes

  • Remove all “X’s house” type nodes, unless X seems a relevant public figure. E.g. has one of the prefixes lawyer - محامي - doctor - دكتور - Mukhtar - مختار, edited by multiple people, etc.

  • If it has a “ديوان” in the name (Roughly translated to Saloon / pub / community meeting area), keep it (Currently looking for the best tag replacement for guest_house).

  • If it seems legitimate but mistagged, just replace the guest_house tag (often with building=yes/aparments). Some typical indicators: اسكان, بناية

Changes so far:

https://www.openstreetmap.org/changeset/50219152

https://www.openstreetmap.org/changeset/50219390

@hadw, Minor note: I think اولا is just a mistyped اولاد, meaning “children”. Probably someone tagged the house of his son/daughter.

If you are guessing the correct main tag, add a FIXME tag that says (in Arabic) something like: “Originally mapped as guest_house, using maps.me. Assumed to be xxxxxx because yyyyyy. Please verify.”

Ok.

I decided to inspect the rest of the middle east, and simple overpass queries reveal a treasure trove of personal bookmarks. Do I have the green-light to exterminate them?

For instance, the following query caught 499 nodes. The vast majority is personal bookmarks.


node[tourism=guest_house]["name"~"منزل"];node[tourism=guest_house]["name"~"دار"];node[tourism=guest_house]["name"~"بيت"];

node[historic=castle]["name"~"منزل"];node[historic=castle]["name"~"دار"];node[historic=castle]["name"~"بيت"];

node[tourism=attraction]["name"~"منزل"];node[tourism=attraction]["name"~"دار"];node[tourism=attraction]["name"~"بيت"];

/* 3 variations of the word home/house */

The location distribution is interesting. Image: https://i.imgur.com/VGIq9nQ.png

Thanks for the analysis!

Unlikely though it is to get a reply, I’d definitely try and contact the mappers concerned first via changeset discussion comments. If you get a case of mappers ignoring comments but keeping adding “home” locations, drop a mail to the DWG and we can send a message that they have to read before continuing mapping. That’ll stop those who have no idea what OSM even is, and at that stage it’s definitely OK to get rid of obviously dodgy MAPS.ME-entered locations (subject to the caveats discussed above about what might be a “useful” entry).

Best Regards,

Andy from the DWG

Is that procedure really necessary? There are probably thousands of personal bookmarks. Contacting them all is time-consuming and inefficient. Additionally, most of them are 1 -time Maps.Me editors. They add their home and disappear. No real need for blocking them, so contacting the DWG will consume the DWG’s time needlessly. Lastly, Maps.ME 1-timers never answer. The chance is 1 in a thousand.

On a tangent: The situation seems severe. Are there any efforts being made to make the MAPS.ME devs fix their app? In its current state, the app is doing so much damage. Simple UI changes / notices may be enough to get people to use the bookmarks rather than the map editor.

You could either contact them via one of the methods on http://maps.me/en/home or write to the OSM talk mailing list which is followed by one of the devs, as you can see in e.g. https://lists.openstreetmap.org/pipermail/talk/2017-July/078273.html

In my view, maps.me is one of the biggest causes of bad data this year.

I think they urgently need to solve the problem of the lack of effective accountability. Unfortunately the only way of doing that would probably be for one of their staff to review changeset comments on their edits that don’t get a response within, say, a week, and I doubt they would have a viable business model if they had to do that.

IMHO, not only added bad data to the database, but also show “bad” data to their users, since they are offering an advertising service to POIs, so, when you search for something, which will came first, the bests items or the sponsored ones?

I looked a bit more into this, and wow. Sorry for straying a bit off topic, but maps.me is slowly destroying the middle east. This is very serious. I assume the experienced mappers of Europe / USA have kept things in check. But there are considerably less dedicated mappers in the middle east. Overpass is not even needed, just browse a major city for a while.

Taking that Baghdad link as an example - there’s nothing else there, and people are filling in the map the only way they know how!

Maybe if someone who spoke fluent Arabic could try and get in touch with them they might be able to convert a small percentage to “mapping properly” :slight_smile:

Also - just in case you’re not aware - you can use http://mmwatch.osmz.ru/?country=Iraq to highlight (and revert) MAPS.ME edits on a country by country basis.

Added to wiki https://wiki.openstreetmap.org/wiki/Quality_assurance#Monitoring_tools Quality Assurance Monitoring Tools

I’ve deleted hundreds of personal houses. But now I’m wondering:

Certain areas of the world have no addresses for various reasons. What if this is helpful for people to navigate, where names are kind of ad-hoc replacements for addresses?

Maybe I’m over-thinking this and it’s just reckless maps.me edits, but I don’t want to reduce the map usability to anyone.

It seems I had at least a single false-positive: http://www.openstreetmap.org/note/1070780

I’ll stop deleting until/if more people from those regions explain things or discuss this.

I will keep deleting the blatantly obvious: My house, My uncle’s house, etc.

I wonder if some kind of “legal amenity” tagging would make sense for some of these - somehow indicating that “here is the location of someone who can mediate in a dispute”. MAPS.ME displays the name for http://www.openstreetmap.org/node/2866177983 (which is an “amenity=solicitors”) and https://www.openstreetmap.org/node/4451574496 (“office=solicitors”). I don’t think that “solicitors” is quite right, but maybe a different one from https://taginfo.openstreetmap.org/keys/office#values might be?