Thanks John. I understand now that this is meant only as a
high-level diagram without much detail.
This is different from MiloÅ's ER diagram which is (more or less)
fully explicit.
M.
On 20/07/2020 11:46, John P. McCrae
wrote:
Hi Michal,
Hi John,
Excuse me if I'm uninformed (I'm new here) but I'm not
sure if I
understand correctly how to read your diagram. I do
understand this:
- The boxes are types.
- Most of the type names are sef-explanatory to me (but
see questions 2,
3, 4, 5 below).
- When a box is inside another another box it's an
"is-a" relation.
Yes, those are subtypesÂ
- The lines between the boxes are relations between
instances of those
types.
They are RDF triples, I did not include property
namesÂ
That much is clear and makes sense to me. But then:
Question 1: Do I assume correctly that the lines are
one-to-many "has-a"
relations, from one "haver" to many "havees" in the
direction of the
arrow?
I didn't put in any cardinality restrictions, there
are some obviouslyÂ
Question 2: What's a 'Lexicographic Component'?
That is our terminology for an element that models
the recursive structure of a dictionaryÂ
Question 3: What's a 'Reference'? (From the diagram,
it's something a
Sense has, and it has a Definition).
A reference is a language-independent reference in a
non-lexicographic resource, such as an ontology,
encyclopedia or knowledge graph.
Question 4: In 'Citation (WIP)', what's "WIP"?
Work In ProgressÂ
Question 5: 'Usage' probably means labels such as
"archaic", "vulgar",
"such-and-such dialect" and so on, right?
Yes, that is what I am referringÂto.
Question 5: 'Morphosyntactic Properties' is in the
diagram twice (two
boxes). Are those two different types (and should they
have different
names then) or are they one and the same type (and
should they be just
one box then)?
They are the same, I just wanted to avoid some
unnecessary arrows.Â
Question 6: What do the background colours of some of
the boxes mean?
The green, blue and purple (if that's what it is).
I was trying to make clear which elements belong to
only the entry (blue), only the sense (purple) or either
(green).
Question 7 (probably my most important one): How do
instances of the
types actually contain data? Is each of them just a
single string (an
Example is a string, a Translation is a string etc.)? Or
does each type
decompose further into something like
attributes/properties/children
which are not shown in the diagram? If it's the former
(= each instance
is a single string) then I'm surprised not to see more
relations between
the boxes; for example, an Example should be allowed to
have
Translations, a Collocation should be allowed to have
Examples. If it's
the latter (= there exist attributes which are not shown
in the diagram)
then I'd be curious to see them; I guess such details
will need to be
made specific in the eventual standard we are going to
produce.
The diagram is quite simplified so I skipped most of
the details here. Some of them can have labels attached
to them, or in some cases, there is more sophisticated
modelling that is fully described in the specification.
The diagram is really only meant to give a high-level
overview of the elements we have in the model, so we can
be clear about a minimal set of requirements for
LEXIDMA.
Regards,
John
These are honest fact-finding questions, by the way. I
just want to be
able to read the diagram as it was intended. The other
diagram, the ER
one tha MiloÅ made, I don't have any lack of clarity
there because I
"speak" ER and because I was involved in the debates
from which it
emerged.
Thanks!
M.
On 2020-06-25 15:04, John P. McCrae wrote:
> Hi all,
>
> I created an updated diagram for the LEXIDMA work
based on the key
> elements of the OntoLex model that are of relevance
to the group.
>
> Some points
>
> 1/ We do allow recursion but only of a single very
specific element
> called the "lexicographic component". This is
mostly due to the way
> that the lexicographic module was added on top of
the core model
> 2/ Most of the elements are allowed on either the
entry or the sense
> except for definitions, usages (which may not be
good), collocations
> (also maybe not quite right), and etymology.
Language is also not
> allowed on forms (it is assumed to be inherited
from the entry).
>
> Regards,
> John
---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the
OASIS TC that
generates this mail. Follow this link to all your TCs
in OASIS at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php
|