Re: Translations for schema.org?

Am 09.11.21 um 22:57 schrieb Dan Brickley:
>
> What did you expect me to say? "You're right, let's translate 2400 schema
> definitions into attempto without evaluating how well it works first"?>

Well, since you ask me, I will answer.

If you'd rather not want to read about what's expected, don't ask next
time. :)


I expect you to give suggestions on this mailing list enough
consideration so as not to grab the most obvious non-point to make,
which is to answer:

> The audience for aircraft service manuals may be relevantly
> different to our audiences.

Obviously, neither Ericsson nor Kodak nor Alcatel build jetfighters, yet
they employ controlled language frameworks to keep their documentation
and translation load at bay.


If you think this is still not close enough to the audience of
Schema.org, may be the audience of
"Semantics_of_Business_Vocabulary_and_Business_Rules" by the Object
Management Group is:

*
https://en.wikipedia.org/wiki/Semantics_of_Business_Vocabulary_and_Business_Rules



Not close enough to the audience of Schema.org?

Maybe Common Logic (ISO/IEC 24707:2007):

* https://en.wikipedia.org/wiki/Common_Logic


Also not close enough to the audience of Schema.org?

Honestly, I do not intend to insult anyone, but I'm not convinced yet
that we, the audience of Schema.org, are that exquisit.


> I'd publish schemas in Toki Pona or Unkleftish Beholding if there was
> general consensus it was an improvement on our current use of English.
> But seriously there are so many ways we could improve on things,
> including work on https://schema.org/docs/styleguide.html )

Oh dude, aren't we funny? Let's throw the names of two exotic language
concepts at a serious argument! May be this will silence the guy and buy
me the hearts of the audience?


You ask me what I expect of you?

Not that, actually, and you know that.

I expect of you, if not for any other reason than at least because you
are here for Google, that you come up with something other than this
kind of trick.


> Let's see how it looks in practice?>
> [...]

Thank you for the quote; I am familiar with the style of definitions
used in the Schema.org vocabulary, because sometimes I have to spent
hours and hours to find my way around, though I am a reader who is
really not faint of heart. Often enough, I will consult with colleagues
who publish scientific papers in machine learning to contemplate on the
wonders and secrets.


No doubt, 2.400 schema definitions are a lot to formalize.

I am sure you are aware of both of the concepts of User Stories and Use
Cases as well as the concept of Continuous Integration.


Ad 1 -- The examples contained in many of the Schema definitions are not
so different from User Stories and Use Cases.

That can be formalized in a controlled language, besides the usual
"don't do this, instead do that" and "if yes, then no".


Ad 2 -- 2.400 schema definitions in "soft style" (that is, unformalized)
is a big Technical Dept, no doubt about that.

Let's start to slow the growth of that dept with a good perspective to
reverse it at some time not so far in the future by replacing one soft
schema definition after another by a formalized definition.

An obvious precedence order recommends itself: Start at the top of the
schema hierarchy and on every following level below, go on with the
schema definition which has the most children, an so on. Multiple
inheritance is not in the way of this concept.

In this manner, continuously replace soft definitions by formalized
definitions.

Note that formalized definitions can still express "softness" or
"vagueness".


Formalizing definitions by means of a controlled language does not mean
to throw OWL at the text of the definitions. That's knowledge
representation.

Rather, employing a controlled language in the text of schema
definitions essentially means two things:


a -- Restrict the grammar and the vocabulary (that is: the set of words)
used in the natural language descriptions in such a way that they can be
marked up more or less automatically. This is essentially the bread and
butter of controlled natural languages.

Here's a motivational example:

If you think it's helpful and easy for Google to parse tables from
websites and from Wikipedia, than the use of controlled natural language
is necessarily helpful and easy, too, because the relationship between
the table concept and the controlled natural language concept is exactly
the same relationship as between Decision Tables and Decision Trees,
which is to say the concepts are semantically congruent (= interchangeable).


b -- Produce natural, but controlled language definitions from formal
definitions already available in the Schema.org vocabulary (e.g., the
prefered, but not strict, inheritance links).


What you get, besides the necessary structure to automatically create
natural language descriptions and translations (!), is presumably a
metacircular definition of the Schema.org vocabulary.

You think you don't need this? There's already a small degree of
metacircular definitions in the Schema.org vocabulary. Growing it will
help to clean up and to stay up to date.


Btw, 2.400 schemas make a nice start for a Semantic Wiki, maybe with
fuzzy links.


Having said all that, I still admit the possibility that controlled
natural language might not be the way to go in this matter.

There may be good reasons not to go this way, though currently none
springs to my mind. :)

But, really, dear Dan:

"Our audience doesn't build aircraft" and "Aren't Toki Pona and
Unkleftish Beholding two funny names?" are not among the good reasons
not to, for sure.


By the way, do you know the novel "A Void" by Georges Perec?

* https://en.wikipedia.org/wiki/A_Void

It's a strongly recommended read for all people who can type "Toki Pona"
and "Unkleftish Beholding". :)


Bests

-- Dan




Am 09.11.21 um 22:57 schrieb Dan Brickley:
> On Tue, 9 Nov 2021 at 21:13, <dan.btown@hotmail.de> wrote
> 
>>
>> Am 09.11.21 um 21:16 schrieb Dan Brickley:
>>> with regard to controlled English and simple English variants I think
>>> they are great as inspirations but it isn't clear that they'll
>> necessarily
>>> be a usability improvement. We could make some experiments though!
>>
>> You are saying that a field with 25 to 45 years of research and practice
>> which, among other things, produces thousands and thousands of pages of
>> safety-relevant service manuals for aircrafts, and which provides
>> companies like IBM, Ericsson, Siemens, Caterpillar, General Motors, Océ,
>> Scania, Dassault, Xerox, Sun, Nortel, Kodak, and Alcatel, to name only a
>> few, with formally well rooted frameworks for documentation and
>> translation thereof is "great as inspiration, but not clear that [it]
>> will necessarily be a usability improvement", the latter of which is one
>> of their reasons-of-being?
>>
> 
> What did you expect me to say? "You're right, let's translate 2400 schema
> definitions into attempto without evaluating how well it works first"?
> 
> The audience for aircraft service manuals may be relevantly different to
> our audiences.
> 
> https://en.wikipedia.org/wiki/Attempto_Controlled_English is very
> interesting, but it seems to be attempting something grander than other
> simple English's in terms of regimented structure, rather than simply a
> restricted vocabulary.
> 
> It reminds me a little of Metalog,
> https://www.w3.org/TandS/QL/QL98/pp/metalog.html
> 
> I'd publish schemas in Toki Pona or Unkleftish Beholding if there was
> general consensus it was an improvement on our current use of English. But
> seriously there are so many ways we could improve on things, including work
> on https://schema.org/docs/styleguide.html )
> 
> Let's see how it looks in practice?
> 
> Care to translate e.g. /ClaimReview "A fact-checking review of claims made
> (or reported) in some creative work (referenced via itemReviewed)."
> 
> Or https://schema.org/OpinionNewsArticle "An OpinionNewsArticle
> <https://schema.org/OpinionNewsArticle> is a NewsArticle
> <https://schema.org/NewsArticle> that primarily expresses opinions rather
> than journalistic reporting of news and events. For example, a NewsArticle
> <https://schema.org/NewsArticle> consisting of a column or Blog
> <https://schema.org/Blog>/BlogPosting <https://schema.org/BlogPosting> entry
> in the Opinions section of a news publication."
> 
> https://schema.org/Offer "An offer to transfer some rights to an item or to
> provide a service — for example, an offer to sell tickets to an event, to
> rent the DVD of a movie, to stream a TV show over the internet, to repair a
> motorcycle, or to loan a book.
> 
> Note: As the businessFunction <https://schema.org/businessFunction> property,
> which identifies the form of offer (e.g. sell, lease, repair, dispose),
> defaults to http://purl.org/goodrelations/v1#Sell; an Offer without a
> defined businessFunction value can be assumed to be an offer to sell.
> 
> For GTIN <http://www.gs1.org/barcodes/technical/idkeys/gtin>-related
> fields, see Check Digit calculator
> <http://www.gs1.org/barcodes/support/check_digit_calculator> and validation
> guide <http://www.gs1us.org/resources/standards/gtin-validation-guide> from
> GS1 <http://www.gs1.org/>."
> 
> Dan
> 
> 
> 
>> Okay.
>>
>> Not so bad you noticed.
>>
>>
>> -- Dan
>>
>>
>

Received on Tuesday, 9 November 2021 23:49:44 UTC