Idea to get better issue requests in the NSI

[GA-Kevin-Codes]

Hello all! I am trying to think of a way to get better issue requests / additions to the NSI by those with little to no knowledge about coding or the projects structure. I have created https://ga-kevin-codes.github.io/Suggest-A-Brand/ as a proof of concept.

Essentially, you see an item like #11268, which is just a wikipedia link, requiring some lifting on the part of contributors and maintainers.

Instead, we can send them to https://ga-kevin-codes.github.io/Suggest-A-Brand/:

Once there, they can follow our wiki instructions to navigate to the Wikidata item of (in this case) Wintrust Financial:

Back on Suggest-A-Brand, you paste the QID into Wikidata, click lookup, then choose amenity=bank and the locations (in this case US states Illinois and Wisconsin) then click "Open Prefilled GitHub Issue"

This will either prompt you to sign in if you haven't already and if you have send you to the new issue screen with the format set for a contributor to verify and link to the correct page on the NSI. All I did here was click "Preview":

Now with very minimal effort a complete newcomer can suggest a brand with correct formatting and useful information. It's not complete by any stretch, its just an idea and proof of concept (repo is here: https://github.com/GA-Kevin-Codes/Suggest-A-Brand) but I would like thoughts on if this would be beneficial to the NSI as a whole or any questions, comments, or concerns you may have. Thanks! Happy Mapping!

[1ec5]

Cool, this is a lot less error-prone than the current approach (usually) of copying an existing entry. Maybe it could also suggest the location set automatically based on the Wikidata item’s country (P17) or service area (P2541)?

[GA-Kevin-Codes]

💡 Great idea! GA-Kevin-Codes/Suggest-A-Brand#1 created!

[1ec5]

We could also streamline the process when a brand or operator doesn’t even have a Wikipedia article or Wikidata item yet. We should propose an entity schema on Wikidata, which would enable Cradle to offer a simple, guided form to fill out instead of needing to create a full-fledged item from scratch.

[GA-Kevin-Codes]

Definitely a great enhancement that can be done! GA-Kevin-Codes/Suggest-A-Brand#2 created!

[RayBB]

I think it's a wonderful idea and I've thought about doing something similar in the past. Two things that would improve the site:

1. Ability to type in a name and it searches for wikidata.
2. Add an option that's like click here to see examples so people can easily find similar entries.

[GA-Kevin-Codes]

Thanks so much @RayBB! I added GA-Kevin-Codes/Suggest-A-Brand#3 to track 1, help would be appreciated if you know of these things!

For 2, I think we can link to the wiki here on the NSI repo, great idea!

[UKChris-osm]

Very well put together 🥇

[GA-Kevin-Codes]

Thank you, @UKChris-osm!

[ShionAmasato]

Here's an example of a new brand registration based on an issue where this tool was actually used.
#11645

My impression after using it was that I was puzzled as to why the "value" of the top-level tag was a free-form entry.

I was vaguely remembering the plural form of "cloth," so I mistakenly wrote "cloths."

I think it would be better to allow users to select from suggestions to prevent mistakes (and then allow free-form input only if the user selects "Not in these").

[GA-Kevin-Codes]

Hey @ShionAmasato! Thanks for using my little tool! The value tag is freeform mainly because it was easier at first than hardcoding values and then getting into debates about proper tagging etc.. There could perhaps be a check that goes "Did you mean ?" that maybe uses TagInfo? What do you think?

[LaoshuBaby]

Looks like a good idea and this may also a hint for editor, "which tag will be better? convenience or shop?", so if it can bake a list of categories trees will be better

(Also, if have a list of available counties and regions?)

This not only can avoid of making typo but also speed the filling.

(But there truly some situation need to fill from freely for example if this means create a issue and suggest a new category)

[LaoshuBaby]

but also speed the filling

by select a choice directly when input some initial letters

but this truly will make the development more complex

[ShionAmasato]

Hmm. I'm having a hard time coming up with a good idea.

Here are a few fragmented thoughts:

• New values ​​are certainly needed and created (for example, "game store" was separated from "shop=toys"), but I think this happens less frequently now that the classification is more mature than before.
• Another approach, when primarily relying on free input, would be to only display a warning (such as "Probably correct: clothes") if the "distance" (e.g., Levenshtein distance) between the known value and the input value is close.