[Tagging] General tagging system problems

[Daniel Koć]

W dniu 13.05.2015 18:24, Bryce Nesbitt napisał(a):
> On Wed, May 13, 2015 at 1:46 AM, Martin Koppenhoefer
> <dieterdreist at gmail.com> wrote:
> 
>> I'm not convinced that such a generic approach will help to get
>> unambiguous tagging, e.g.
>> children + education + building (= school building) ->
>> architectural school? language school? ...
>> vehicle + education (= driving school) -> self learning autonomous
>> vehicles? school bus? a bus that offers educational services? ...
>> area + tree (= forest/wood) -> a big tree? an area foreseen for
>> trees? an orchard? ...
>> building + sleep (= hotel/hostel/...) -> a motel? a dormitory? a
>> matress factory? ...
> 
> +1 well illustrated

Oh, so it have sounded like I disagree with current ability to go down 
into details? Sorry for that! =} My intention was not to get rid of them 
and have _only_ the basics, but to be able to have _better_ (=more 
general/universal/"basic") basics.

I have to admit: we're damn good at details, but it's much easier to 
extend them than to upgrade the generalities. And that's why we can 
describe very complex properties, but we're unable to use proper 
categories with general/specific relations between the objects.

For example we know the natural=forest and natural=wood have something 
in common, but we have no tools to show how they relate to each other 
(natural=* is just too broad) nor to let mappers describe general tree 
area when needed (natural=trees + area=yes alone would do, but that's 
another "case" to remember with different wording, not resembling 
"forest" nor "wood").

> The hierarchy approach sidesteps some of that
> 
>> building
>> 
>> what type of building?

BTW: building is actually a great example of right basics we already 
have. It is general enough to say building=yes and it's useful, but we 
may narrow it down if needed - so it works as advertised. =}

> But I think it better to go the other way.  Start with the "duck":
> 
>> driving_school

>            driving school  + private_vehicle + fee =   things open to
> the general public, services, things you book in advance
>            driving school  +  heavy goods vehicle+ fee =   services,
> specialty services
> 
> Which other than the categories, pretty much is the wiki of today.

OK, but driving school is not the most basic object around we can find 
(even the name reveals that it's a complex object!), so we still have no 
tools for proper categorization from the top. We have to start from the 
middle (keys like natural, amenity etc) and we need a lot of objects.

In other words, this example doesn't change anything, that's what we 
have now: we still rule at details and suck at generics.

***

Just today I've learned about linguistic theory called "natural semantic 
metalanguage" (NSM) and this is exactly what I think we need to make the 
tagging system sane and easy to navigate. The "bricks" will be 
different, because we focus on geospatial informations only, but the 
rule is the same - they have to be as elementary as it gets:

"An explication is a breakdown of a non-prime concept into prime ones.

E.g., Someone X killed someone Y:
someone X did something to someone else Y
because of this, something happened to Y at the same time
because of this, something happened to Y's body
because of this, after this Y was not living anymore"

[ 
https://en.wikipedia.org/wiki/Natural_semantic_metalanguage#Explication 
]

This form is probably not very useful for small talk =} , but it has 
some big advantages: we need only tiny vocabulary to tell everything and 
the objects do not overlap.

And this is a problem with existing tag set and Wiki today - many of our 
basics are not that basic really and the vocabulary is so big, that we 
have to check a lot of things. Let alone casual mappers...

***

I hope this time my message will be more clear:

1. Everything is well with how we describe detailed properties and I see 
no need to touch it.

2. At the same time we have serious problems with general objects 
classification rules (incompleteness, overlapping) and generic tag 
overload.

3. That result is once you have the right tagging scheme, you're safe, 
but finding it can be hard or impossible. Another consequence: Wiki is 
no longer just a helpful documentation, but the only way to manage 
everything at all, and so its' importance is overstated and 
misunderstood (hence countless discussions about "approved" tag schemes 
as if it's "official" somehow).

4. We can still use complex ideas like "supermarket" or "driving 
school", because they're shorter way of describing things, but if they 
prove to be too complex, we have tools to express it anyway - easy and 
according to existing rules.

-- 
"The train is always on time / The trick is to be ready to put your bags 
down" [A. Cohen]