qwantmaps
qwantmaps copied to clipboard
Error in search name - Isolated dwelling
It's a very specific case (I haven't looked for any other):
In Qwant Maps, I can't find the isolated dwelling name "Le Petit Rocher", located in "Pacé" (zipcode 35740), France by searching "Le Petit Rocher, 35740 Pacé".
In order to find it, I have to search for "5031 Le Petit Rocher":
However, it has the right name and attributes in OSM, as you can see:
Steps to reproduce:
- Search for "Le Petit Rocher, 35740 Pacé" in Qwant Maps --> no correct results
- Search for "5031 Le Petit Rocher"--> correct result found
Hope this helps ! Talw3g
I have seen this with other buildings names as well. It seems from the readme that QwantMaps syncs data from OSM on a monthly basis but this still happens with buildings that have been updated on OSM over a month ago. Does building information not get used?
I've updated the information of this specific place a few month ago, so I don't think it has to do with QwantMaps not being up-to-date with OSM data... I believe there is indeed a sync lag, but for what I've seen it's not more than a few days.
It looks like QwantMaps just doesn't include some Building name data. Possibly related to the reduced schema sized mentioned here?
https://github.com/QwantResearch/kartotherian_docker/issues/67
I'm really no expert on this, but I don't think it has to do with missing data : the name of the dwelling ("Le Petit Rocher") does appear in QM and is a part of the search results... The "5031" before the name is a real mystery though.
Anyone from Qwant able to comment on this?
The database used to search on Qwant Maps contains only a selected subset of OSM data.
In this specific case, objects with place=isolated_dwelling
or building names are not currently indexed by the search engine.
Interestingly "5031 Le Petit Rocher" exists in our database as an address because it's part of BANO, the data source we are using for french addresses. I suppose this might be because this is how the dwelling is referenced in the Cadastre, for example.
We are working on extending the database used by the geographical search engine to cover more diverse items from OSM data. This comes with many trade-offs about performance and relevance of the results suggestions, as we rely a lot on auto-completion to provide a good user experience.
Objects with place=isolated_dwelling
are especially challenging in this context, as they are numerous minor items that often have very common names, and they could easily be confused with other points of interest. In France alone for example, there are
198 isolated_dwelling
places named "Le Moulin", 145 named "Le Château", or 121 named "La Gare".
Anyway, even if this will require extra care, I agree they should be included in our results.