[OSM-dev] Fwd: Name suggestions for iD

Aaron Lidman aaronlidman at gmail.com
Sat Nov 2 19:14:15 UTC 2013


Yes Serge, seems like I was thinking of a different issue than you 
brought up.

Setting other tags based on a name selection from the preset bar is very 
interesting. The current behavior (category -> details) should stay as 
the most preferred (recommended) way of doing it but if preset search 
results aren't good enough we could fall back to suggesting close name 
matches and pre-fill known common details.

One immediate issue I see is that it might go against expectations and 
confuse a user, like where did all this data come from? Right now they 
select a category and one or two tags are added, which is pretty simple 
to deduce. If we 'quick add' a bunch of details lots happens in a short 
amount of time, which might be confusing. Maybe we can prompt 
expectations with some kind of icon, message, or animation in some way, 
I'm not sure.

Seeing as how the preset model used in iD already different from how 
other editors work I'm more open to trying this out. Technically, I 
don't think it would be too difficult, my bigger concern is about user 
expectations and reasonable defaults. The name-suggestion-index is 
already almost able to do this, we could simply rename the 'translation' 
property to something like 'details' or 'prefill' and use whatever 
values are in there. We could also keep the basic implementation of name 
suggestions (category -> details -> type name -> suggest name and only 
fill in the name value) for users who want more control over the process.

I'm going to look into the specific details of how preset suggestions 
work and get John's thoughts on this. I see the value is doing this I'm 
just not sure about prioritizing it right now with the inevitable unseen 
technical issues and community push back. Basic name suggestions are 
already done in iD (https://github.com/systemed/iD/compare/name-suggest) 
and this name stuff has already been a pretty big tangent so I might 
have to push this back to later.

Thanks for your thoughts, this has been helpful.

On 11/2/13 5:15 AM, Serge Wroclawski wrote:
>   Aaron,
>
> I think you're conflating two separate issues, which is why things
> seem more intimidating than they are.
>
> I'm going to leave out any discussion about changing existing data-
> since that's where things would get controversial quickly.
>
> Instead, I'm going to focus on tagging issues in the editor.
>
> In iD, when you create an object, you can use the search bar to create
> an object with tags from a set of templates.
>
> Name expansion is good, it's really good.
>
> But while name=McDonald's is good.
>
> name (or brand)=McDonalds
> amenity=fast_food
> cuisine=burgers
>
> is great. And it's great for a few reasons
>
> 1. It makes it easy to add new information about a feature without too
> much thinking about it. Right now, iD's search functionality makes
> adding a new feature easily, but you have to know the category first.
>
> If we have features fully searchable by name, we eliminate a step (category).
>
> 2. It reduces the chances for making a tagging mistake
>
> Right now, I see people choosing the wrong category for features.
> restaurant=yes, name=McDonalds.
>
> Why do people make this error? Because they aren't looking at all the
> categories and thinking about them, they just pick one quickly.  And
> for some people, this may be their first edit, so we can certainly
> understand how they might want to just pick *something*.
>
> 3. It provides built in examples.
>
> Everyone learns differently, but I learn by example. If I knew that
> McDonalds is a fast-food, then I might ask "What's a Ruth Chris?" and
> play with that.
>
> 4. It helps emphasize tags we want filled in
>
> Some McDonalds have free wifi, others don't. If that's something we
> care about (just as an example) then we can add it to the preset, thus
> encouraging users to collect that information.
>
> - Serge
>





More information about the dev mailing list