Asia - ok, comes from ‘is_in:continent’ tag of Israel Israel - ok, except for some strange names. Who is Милан Јелисавчић? Tel Aviv District - comes from the hard work of dimka and talkat - cheers! But this is a relation, without a ‘place’ or ‘is_in’ tag - so how is it a part of this description? Ilesha’ Kebots - was a node with ‘place=locality’, I changed it to ‘place=village’. Should not appear in searches soon. West Bank is a node with ‘place=county’ - should it be? Probably not related as in Tel Aviv District. Tel Aviv-Yafo - ok, but how come it is in the west bank? It has its own ‘is_in:continent’ and ‘is_in:country’ - should it have just the latter?
So, how do we bring order here? What is the right hierarchy for Israel? And what is up with the is_in tag?!
Beside the fact that its a little bit messed up there it is also missing the information “central israel” i think.
If you search in eilat you will have also the westbank in it. Very strange.
The search engine is Nominatim.
Its search hierarchy is described here.
Correct, Asia has the highest value.
He’s a mapper who added name:sr and official_name:sr.
According to ISO 639-1.
It’s part of Nominatim’s hierarchy calculation.
This is the closest feature on Nominatim’s hierarchy.
Actually, it’s a “Mechina Kdam Zvait”. Fixed now
I would say that this is a bug in Nominatim.
Since Tel Aviv is a place=city, it has rank of 16.
How come it’s “in” a feature of rank 20? (place=locality is rank 20…)
It’s the closest county (Rank 12)
Follow the ranks as described here and all should be as clear as muddy water…
Thanks for the details - I now have more questions
Nominatim makes use of the place key without mentioning it on its wiki. Moreover, it uses values that are not documented for the place key, such as localities, counties and regions. Is this another bug in Nominatim?
The localities, counties and regions come from a huge import of the west bank. Changing them to villages or the like seems like an edit war - I really don’t want to go there. But the county and region tags are not documented with ‘place’, and need to be changed to real administrative boundaries. What is the right way to update this data?
Tel Aviv District has admin_level of 8 (right?), so according to the calculation, it is taken at rank 16 (8*2), which is less important than a county (rank 12) or region (rank 10). So how come the dirstrict ‘includes’ the county ‘West Bank’ in Tel Aviv’s search, and not vice versa? Another bug?
The JumpStart import had many bugs, and the data was entered by mappers who were new to OSM, with noone going over their edits.
It’s a long process of cleaning all the data that was imported.
Still, I believe that the benefits of having this data are greater than its cleanliness drawbacks.
I would go over all the place=region and see whether they really are regions, starting with these nodes:
Om Al Toot, node 432403462
Bir Ziet rafugee camp, node 432285689
Alteerah, node 432293792
Dahyat Alrehan, node 432278682
Al Sakar Region, node 432295516
Baten al hawa area, node 432296399
Aen Munjad, node 432269952
Ramallah Altahtah, node 432282318
Ein Musbah, node 432277882
AL Tarbe’, node 432264920
Dahiat Almazra’ah, node 432297190
Al Amriyah, node 432097524
Albaloo’, node 432267280
Almasion, node 432277980
Rafidia Al Balad, node 432101379
Alshurafa, node 432261625
Al Makhfiyah, node 432171818
Sateh Marhabah, node 432269101
Beduin, node 432276470
Velal Aldahyeh, node 432267623
Al Ein Camp, node 432041254
Ahmad Yaseen Region, node 432280444
Old City, node 431933570
Al Dahiah, node 432006141
'Anza, node 431787354
Balata Camp, node 432140467
Employee Area, node 431948338
Industrial Area, node 432178148
Industrial Area, node 432179070
Al Syam, node 432261199
Al Zawiya, node 432544937
Al-Masakin, node 431920714
Industrial Area, node 432282309
Haret 'l Noor, node 432289238
Al Wa’rat, node 432384362
Also, I guess we should add place=region in the center of all Israel’s 6 districts.
(Maybe also add place=county in the center of all municipalities?)
You have very good questions!
I pointed the user who maintains Nominatim to this thread and these questions, and I hope he would have valuable insights we could implement.
Without getting too political, how should Israeli nodes be defined to overcome the influence of these nodes?
talkat, how about your suggestion to add region nodes in Israeli districts?
Seems that such changes take a while to reflect on Nominatim…
From what I saw thus far, no such artifacts seem to appear on most of the locations inside Israel (there are problems in West Bank/Judea and Samaria which I hope can be fixed by manual tweak of the local copy of the database.)
I would ask everybody from now on to check the above Nominatim copy and report whatever problems here as well.
Abrukh is somehow right that we should also identifier the issues of the global Nominatim an address them.
There is no benifit for the most users if we tweak our DB to display correct results , but all users of OSm.org will continue to see those issues.
I mean it would be very clear that this place tag would belong to the country (admin_level = 2).
For me personally it doesn’t make sense that a node placed in another country is influencing a search of the country nearby.
BTW: We have this issue all over Israel. It seems like the place_region tag is seen all over the place.
But I can also see that adrukh has fixed a lot if not all of them.
Correct (I followed the installation instructions “to the letter”).
Of course if all the problems could be ultimately solved with the “is_in” tag or some other “legal” tagging, I’m in for it. Our localized Nominatim would just be updated more frequently, that’s all.
