OSM: Missing geometry for Bekasi in Indonesia

[ktk]

I'm trying to get all streets in the Jakarta area called Jabodetabek 1, Wikidata ID Q1015978.

So I'm querying Jabodetabek from Wikidata and then try to get everything "contained":

https://qlever.cs.uni-freiburg.de/osm-planet/aAjujJ

You can see that Bekasi is not part of it, which it should be. I do get it in Wikidata:

https://qlever.dev/wikidata/zSnFk7

So I check out if the Wikidata-ID is attached in OSM and it looks like it:

https://www.openstreetmap.org/node/318882209#map=19/-6.234986/106.994544

I see the Q-ID attached.

When I set the object of the Q-id in the query, I do get something back:

https://qlever.cs.uni-freiburg.de/osm-planet/sRkPz4

But as soon as I enable line 15 with the sfContains, I get zero results.

If I compare that with Bogor for example, the only obvious difference I see is that Bekasi is not of type=boundary. Is this the problem and then sfContains is not calculated in the osm2rdf process?

[hannahbast]

@ktk If you just look at the regions returned by the Wikidata query, you see that Bekasi is missing:

https://qlever.cs.uni-freiburg.de/osm-planet/Af84Pp

(If you go to the map view, you can click inside each of the areas, and there is nothing if you click somewhere inside Bekasi.)

[hannahbast]

And the region from OSM linked to the Bekasi in Wikidata is this: https://www.openstreetmap.org/relation/14765575, which does not contain the city of Bekasi

[ktk]

ok interesting, I don't think the shape https://www.openstreetmap.org/relation/14765575 is correct for Bekasi. Will try to find some official shapes or people that have a clue. I think the explanation is then that there is no correct Bekasi shape in OSM at the moment.

Bekasi on google maps looks exactly like what is empty on my first query

[ktk]

Ok there is Bekasi: https://www.wikidata.org/wiki/Q10392
And Bekasi Regency, which explains the way larger shape: https://www.wikidata.org/wiki/Q10340

Will have to see where Q10392 is missing

[ktk]

The correct shape would be https://www.openstreetmap.org/relation/14509733. I just don't understand yet where it is missing. Will investigate tomorrow.

[hannahbast]

@ktk The geometry of https://www.openstreetmap.org/relation/14509733 was broken in the OSM input data, hence there is nogeo:hasGeometry triple for it.

According to OSM, this has been fixed 13 days ago: https://www.openstreetmap.org/changeset/163718463 , just after the respective OSM dump was published. I am in the process of building an index for the latest OSM dump (available since this morning), which should fix the problem.

These data problems in OSM are not that infrequent. However, they typically do not affect larger regions.

[ktk]

Ah I did see the change but I didn't make that connection. Looking forward to the new version!

[hannahbast]

@ktk After the update to the next version of the dataset, the Bekasi-hole has disappeared: https://qlever.cs.uni-freiburg.de/osm-planet/aAjujJ

[ktk]

Fantastic, thanks!