Good afternoon dear mappers,
HellMap recently pointed me to the NSI project for "Canonical features". Released version of the list here.
I already added parcel locker for Latvijas Pasts and Venipak (waiting for "release" for them to become available while mapping).
Also, am waiting for approval for Omniva lockers to be extended to LV and LT. Not sure why addition of new brands was quicker than modification of existing one.
Do we have somebody familiar with the process here?
Also, if somebody has some ideas of what brands should be added into that list, I'm willing to try and add them.
If I understand correctly, acceptance criteria is having at least 50 points worldwide.
Full disclosure, I'm not familiar with editing wikidata, because I used only existing instances. But how hard can it be, right?)
I see that chats around here are all in Latvian, but I'm only learning it, so I hope you'll bear with me.
Please fill free to answer in any language, it's that I just can't properly reply for now.
Hey! English is fine, I think most people here speak English.
Here's a parsed brand list from Latvia that I made some time ago - it has a bunch of brands that do and do not have NSI entries - https://osmlatvija.github.io/Osmalyzer/Common%20Brands%20report.html
I think there's nothing more to the process beyond what you already did - submit issue or PR to the NSI repo
One thing I'll note based on that parsed list is to also consider the incorrect spellings and add those for auto-detect/correct
Wow, very extensive list. Will try to go through it.
By the way, where did you see the 50 entry threshold?
HellMap said:
I think there's nothing more to the process beyond what you already did - submit issue or PR to the NSI repo
I thought that maybe there was some required step for modifications that I missed.
Or in the very best case scenario - some maintainer for NSI would be here, and would approve my PR :blush:
HellMap said:
One thing I'll note based on that parsed list is to also consider the incorrect spellings and add those for auto-detect/correct
NSI has some form of auto-detection. At least for different cases. But if we have variation in naming, like I recently found some pakomats having name "Pasta stacija" (or close), it can be added to alt_names.
Will pay attention in future.
HellMap said:
By the way, where did you see the 50 entry threshold?
Ohh, sorry it was 30. Probably confused it with 50 that is mention later in that text.
https://github.com/osmlab/name-suggestion-index/wiki/Judge-Case#notability
At least that's assuming Latvia is not a "OSM data shortage region"
Ah, I see, and they are also talking about exisiting POIs already tagged not just theoretical minimum
I have no idea how many of these brands in Latvia actually have Wikidata entries. I assume many do.
No, I think this applies to "theoretical" locations too. Based on the wording for "shortage" regions.
The first thing I search "Mēness aptieka" and its already weird https://www.wikidata.org/wiki/Q57583051
Why?
Their site says this:
Aptieku uzņēmuma „Mēness aptieka” zīmola īpašnieks ir AS „Sentor Farm aptiekas”,
Yes, but I mean, Wikidata label should be the common name and the company/official/legal name in the alt names and the appropriate field(s)
This is one of the reasons why things in NSI get stuck unmerged, when this sort of stuff is left unfixed in Wikidata, and the maintainers have a hard time understanding what is what. So something to keep in mind. I know I've been putting off doing this because there is just so much to fix...
Well, I'll try to go through stuff in your list, and hopefully as a result there will be less much so what needs a fix)
By the way, Mēness aptieka is already in the list, and name is shown as expected (because NSI's name is in priority to wikidata?)
image.png
Yeah, the NSI values and thus iD values use the name set when they made the entry - iD doesn't actually connect to Wikidata to get the values.
Oh, and just to be clear - that list doesn't know anything about what is or isn't added to NSI. We already have a few things added to NSI. I was just using that one as an example because I wanted to see what Wikidata entries are like. The list is simply sorted by the number of mapped occurrences of a certain name. It's meant more for fixing naming inconsistencies. It has a "second" list below that doesn't havbe any names diverging from expected.
HellMap said:
Yeah, the NSI values and thus iD values use the name set when they made the entry - iD doesn't actually connect to Wikidata to get the values.
Is it?
I thought it does, because otherwise I don't understand where iD gets info about post offices and post boxes:
image.png
I understood that your list is not checking this. I just wanted to show that your issue with wikidata for Mēness does not affect mappers.
It gets values from here apparently https://cdn.jsdelivr.net/npm/name-suggestion-index@6.0/dist/presets/nsi-id-presets.min.json
It does so once when you open the editor. I am not aware of it ever connecting to Wikidata directly though.
markalex2209 said:
I understood that your list is not checking this. I just wanted to show that your issue with wikidata for Mēness does not affect mappers.
Yeah, it doesn't affect mappers. It may affect volunteers at the NSI repo having the time or desire to understand and approve the issue. In the ideal world, they would just process everything. But from experience, I can tell you that clear and clean issues get merged and approved MUCH faster. It's not a guarantee, of course. Just something to consider if you were to submit a lot of new issues.
HellMap said:
It does so once when you open the editor. I am not aware of it ever connecting to Wikidata directly though.
Hmm, I don't understand why it's not showing up in the list itself then. Need to dig a bit before proceeding
Sorry, what is not showing up?
Offices and boxes of Latvijas Pasts: https://nsi.guide/index.html?t=brands&tt=pasts
Aha, because it's in operators tree, not brands. Very interesting.
markalex2209 said:
...
Also, am waiting for approval for Omniva lockers to be extended to LV and LT. Not sure why addition of new brands was quicker than modification of existing one.
...
I see that chats around here are all in Latvian, but I'm only learning it, so I hope you'll bear with me.
Please fill free to answer in any language, it's that I just can't properly reply for now.
Hi and welcome - noticed your quality map changes, including note cleanup :)
Most NSI topics seem to be resolved - just a quick note on having cross-country (ha) entries: sometimes there are differences, which would require a separate entry. Brands might be different, maybe there are separate Wikidata entries for separate legal companies etc. Not saying it's the case with Omniva, but worth doublechecking.
For example, Aibe operates in Latvia and Lithuania, but in Lithuania it's "Aibė". That dot on top of "e" might seem insignificant for NSI maintainers from another continent :)
Regarding the language, English is perfectly fine, and if you are learning Latvian, please feel free to use this as a chance to practice as you please. Same as with OSM editing, there is no shame in making mistakes at all, we all learn by making mistakes.
If somebody from the local community would raise a point of struggling with English, the general guideline would call for switching to Latvian (via translation services or otherwise), but so far it seems to be a self-selection of sufficient knowledge of English.
Thank you for your kind words, Rihards)
Seeing that even DPD with its great deal of recognition causes troubles in tagging, I'll try to be more cautious with such cases.
Regarding language, I'd say my level is probably A1. So there is no actual way for me to convey anything at the moment: I simply have to fall back to English to say anything.
Though I'm happy to try and read more in Latvian, and catch some familiar (easier) words =))
I have a question regarding Depozīta punkts.
Right now they are added as two types: amenity=recycling or amenity=vending_machine
https://github.com/osmlab/name-suggestion-index/commit/df7f4035e802a3145ce07d062a1d92b14b700393
I can only assume that they were added just because they started appearing in this two types, and author didn't actually know that these are the same(?) things.
Should we fix this, and leave only one (and which one)?
Or should we leave it as is (and maybe ascribe some mnemonic meaning, like vending machine is when it's inside a shop, and recycling centre is a standalone building)?
Also, is there any kind of established practice for tagging manual deposit points?
markalex2209 said:
Thank you for your kind words, Rihards)
Seeing that even DPD with its great deal of recognition causes troubles in tagging, I'll try to be more cautious with such cases.
On the other hand, also don't let it dissuade you from making changes - while extra caution is advised, getting blocked is not a good outcome anyway.
Continuous improvement over trying to get it perfect right away :)
4 messages were moved from this topic to #general > depozīta punkti by Rihards Olups.
Omniva was finally merged, and still no release)
I assume this BITĖ and our Bite, while are related, not the same exactly, as in Rihards' example about Aibė?
Yes, it appears Latvian Bite would be Q89915656
Oh great, many thanks!
Link to file to track progress with NSI, if somebody interested: https://1drv.ms/x/s!Arqm2vwx2Fg6uCvg4q7DmfmfuWzj?e=6hgxXH
File is editable, so fill free to add anything useful if I missed it.
List is created from Common brands of OSMalyzer.
Nice! I'm not entirely sure which entries are pending and which exist in NSI? Could we add a status column or something like that?
NSI column is for this. It's just that this column is filtered.
Removed filtration
Oh, I'm blind, sorry :D
Yeah, I saw some rows were collapsed, wasn't sure what that means
No problem)
Oh... wait. Am I not blind? What is "filtered"?
Yes it was. On the header of the column.
It's possible that right now filtration is jumping back and forth, because I'm trying to figure out how to apply filtration only for me.
HellMap said:
Oh... wait. Am I not blind? What is "filtered"?
Oh my, I'm blind too(
I read "Was it" instead of "what is".
On the headers of columns there is a dropdown, where you can select values to be show/hidden from view.
Hmm: https://github.com/osmlab/name-suggestion-index/pull/9553#issuecomment-2132292071
Note that the items without
brand:wikidata
won't be converted to presets until that tag is added later.
I wasn't aware of that. And adding item to wikidata doesn't look like something I'm ready to commit to.
I can help add and fix some stuff for Wikidata since I've edited there before... although I need to make yet another account
HellMap said:
I can help add and fix some stuff for Wikidata since I've edited there before... although I need to make yet another account
It will be great. Please add ids to Excel doc once they exist.
And if you know of really simple guide for how to add brand/business there, I'd appreciate it too.
NSI was finally updated :tada:
Last updated: Dec 22 2024 at 03:46 UTC