improper usage of ana attibute

[matyaskopp]

I think that not all @ana values should be in @ana attribute in the current version of our data.

As I see it - @ana attribute should contain mainly references to taxonomies or features. Our @ana values contain also references to events

ParlaMint/ParlaMint-SI/ParlaMint-SI.xml

Lines 435 to 438 in 3c8ad8a

	<event xml:id="DZ.7" from="2014-08-01" to="2018-06-21">
	<label xml:lang="sl">7. mandat</label>
	<label xml:lang="en">Term 7</label>
	</event>

ParlaMint/ParlaMint-SI/ParlaMint-SI.xml

Lines 653 to 657 in 3c8ad8a

	<relation name="coalition"
	mutual="#party.SMC.1 #party.DeSUS #party.SD"
	from="2014-09-18"
	to="2018-03-14"
	ana="#DZ.7"/>

(and probably some other cases)

@ana should be replaced with attribute ie. @corresp, @synch,...

[TomazErjavec]

@matyaskopp, thanks for referencing this very old but very important issue. Going back in history, in Parla-CLARIN Andrej and me wanted to minimise the number of pointing attributes used, as TEI has a large number of them, often with unclear disctinction in semantics, esp. when you want to apply them to concrete situations.
We decided on @ana as the main pointin attribute, because it has general definition "associating specific analyses or interpretations with appropriate portions of a text" and because it allows "1–∞ occurrences of teidata.pointer](https://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-teidata.pointer.html)" it is also possible to pack several pointers in the same attribute.

But, yes, we also allowed the use of @corresp because TEI often explicitly uses it on some elements, it is also global with up to ∞ pointers and its semantics is relativelly clear and different from @ana, i.e. it is used not for analysis / classification but for saying that the element in some sense corresponds to another element, like in cases of having two transcriptions or that a page break corresponds to a audio file. (But now that I carefully look at cases of @corresp in Parla-CLARIN I see we sometimes used it also for other things, which was a mistake that I have to correct, cf. Parla-CLARIN issue 24).

And then there is @ref, which we used only on placeName (as does TEI] and on affiliation, which is maybe more conentious. The definition is "provides an explicit means of locating a full definition or identity for the entity being named by means of one or more URIs. " and this is implicitly taken to refer to the (text) content of the element with @ref, which is why I think it is not absolutely clear if it belongs to affiliation.

In ParlaMint various corpora use these 3 attributes in different way, at least for their optional data, like ministries, and I agree that some order must be imposed. But taking into account that every change impacts "everything", up to all the conversion scripts.

So, first, we would have to give these 3 good ParlaMint-specific definitions (which are subsets of the TEI ones) and examples and add this to Parla-CLARIN but in ParlaMint make a new subsection 3.5 (starting with the text "The analysis attribute, ana which is also used by many other elements, so we introduce it here at more length: " in 3.4.

@matyaskopp, do you agree with this so far?

If yes, we then look at all cases where any of these 3 used used in ParlaMint, and decide whether to leave the info there, or move it (possibly only some values, where there are many) to other attributes. And then document this in fixings, make the XSLT v2 to v3, change the guidelines, and the XSLT scripts. Which is lots of work, but I guess it is now or never, if we want to start closing issues, and we do.