Open nielsklazenga opened 3 years ago
nielsklazenga
commented
3 years ago See discussion in issue tdwg/tnc#45.
@camwebb , would you agree that intersects includes the TCS 1 "synonym" taxon relationship type, plus relationships with nominal or informal taxon concepts with the same taxon name as the subject taxon concept (and cases where the concept alignment has not been done yet)?
nfranz
commented
3 years ago Possible add, to explain and also show that this is merely or mostly a shorthand: intersects = {is congruent with OR includes OR is included in OR overlaps with} = not {excludes}.
camwebb
commented
3 years ago @nielsklazenga I think it's important to separate the nomenclatural terms (synonymOf) from TC relationship terms. A synonymOf B relates two names, not TCs. I know TCS1 used synonym as (defunct) TC relationship term, but this is one of very few places where I disagree with the TCS1 authors. In our TC mapping work for the Alaska flora we have two fields: TC relationship and synonym type.
nielsklazenga
commented
3 years ago @camwebb, you and me both. Synonyms have the same relationship to a Taxon Concept as the accepted name (taxonName).
I was merely wondering in what circumstances intersects would be used. I was thinking of old literature reports where you do not have reason to doubt the determination, but where the only information you have is the name, which is either the accepted name or a synonym of a Taxon Concept you currently recognise. The name will have to have had a rather checkered usage history.
deepreef
commented
3 years ago I agree completely. The concept of "synonym" is nomenclatural. However, it does play a role in taxonomy, in that assertions of heterotypic synonymy impact taxon concept circumscriptions differently than assertions of homotypic synonymy.
But in any case, I think TCS2 needs to capture "synonymOf" at the granularity of individual TNUs (not names -- if names are maintained as separate entities from TNUs).
camwebb
commented
3 years ago @nielsklazenga Ah yes, I see what you were asking now. And yes, where no TC mapping has been done, but we do know that there is at least on specimen in common between two TCs (which we might indeed know from a synonym nomenclatural statement) at a minimum we can say that two TCs intersect. We find we use this term in a lot. I agree with @nfranz's suggestion of adding an (A OR B OR ...) definition too, though perhaps 'intersects is a superclass of...' is maybe a bit clearer?
nielsklazenga
commented
3 years ago @deepreef, I agree with you too. It is the Taxon Concepts that have synonyms (at least the non-homotypic ones), not the Taxon Names. Your "synonymOf" is the same thing as acceptedNameUsage (#8), which I propose to name taxonConcept, so it can be used more generally, e.g. for vernacular names. (I thought I had already done that, but I guess not). I would also like a synonyms property on the Taxon Concept (TNU), so people can embed synonyms. BTW, as soon as you add an ID to the synonym assertion, the synonym itself is also a TNU.
@camwebb, I actually think @nfranz's formulas are the clearest, but we might just describe them in the usage notes and have a table somewhere with formulas that illustrate the relationships between all taxon relationship types. I think we should avoid any talk of superclasses here though.
nielsklazenga
commented
3 years ago One thing that came up in the discussion in the TNC IG meeting this morning is that people want another label for this term. Easier said than done, but we should think about it. I think the problem might be more with the term overlaps (#55 ), as that could mean essentially the same thing as 'intersects'. I think that 'intersects' is the right label for this term, as sets that are not disjoint intersect and that is what we are trying to say here. Maybe it is better to change 'overlaps' to something like 'partially overlaps'? 'Partially overlapping' is the RCC equivalent of our overlaps.
camwebb
commented
3 years ago Agree @nielsklazenga, the problematic term is overlap. Not really sure 'partially' is the right modifier (objectively), but since RCC-5 uses it, that may be the best way to go.
deepreef
commented
3 years ago The part that's a little confusing to me is that presumably the five standard relationships are mutually exclusive with respect to each other. But "Intersects" could be the same as Congruent, or could be the same as Includes/IncludedIn, or could be the same as Overlaps (but it's definitely at least one of those). In other words, Intersects would be the only one that is non-mutually-exclusive with the others (except Excludes), and the only one that is ambiguous in its logical representation.
Wouldn't an alternative be to keep the five standard ones as they are, then add an optional "not" parameter that could be applied to "Excludes" to represent the same thing as "Intercepts"? I can imagine some value in applying "not" to others as well (e.g., not congruent, without being sure whether it's Includes/IncludedIn or Overlaps).
nielsklazenga
commented
3 years ago @camwebb , yes, I would have thought 'partly', but the meaning is almost exactly the same, which is why I just followed RCC-5.
@deepreef, I see your point. We can make this clear by using a taxonomy:
|- intersects
| |- is congruent with
| |- has proper part
| |- is proper part of
| |- partially overlaps
|- is disjoint toWe can do that in SKOS (which TDWG controlled vocabularies use) using skos:broader (Steve's scripts already take this into account). The problem might be that we want people to use intersects might see this as a divide between intersects and is disjoint to, while we really want them to use the other four terms. So, what we really want to have might be something like this:
|- is congruent with
|- has proper part
|- is proper part of
|- partially overlaps
|- any of the above (intersects)
|- none of the above (is disjoint to)I think that is exactly the same. We just have to find some other way to make clear which the more important terms are.
If we had a 'NOT' parameter, we obviously would not need intersects anymore, but that would break data exchange, as systems could ignore this parameter (we can obviously not make it required) and not exchange it, but still exchange the relationship type (which is required). If we want to use negation, negated terms need to be added to the controlled vocabulary, as has been done for e.g. notatype (that is actually the only example I can think of).
We've got a clear use case for including intersects. If such a use case comes up for 'not congruent', we can consider that as well.
Also, 'not disjoint' (or 'not excluded') feels like a double negative to me.
nielsklazenga
commented
1 year ago @camwebb: Could you provide an example of the use of this term from your work for the doc?
intersects (property)
The taxonomic meanings of the subject and object taxon concepts intersect, i.e. they have at least one member in common.
intersectscan be used as a property on a Taxon Concept object, or as the value of themappingRelationproperty on a Taxon Concept Mapping object. In both cases both the subject and object are Taxon Concepts.Comments
intersectsis the opposite ofisDisjointFromand the union ofisCongruentWith,includes,isIncludedInandpartiallyOverlaps, meaning it can be any of these relations. This relation can be used when the more precise nature of the relationship is not known.Quasi-nomenclatural statements like 'pro parte synonym', 'partial synonym' and 'misapplication', are Taxon Concept Mappings, no matter how imperfect, and, in TCS, are best dealt with using the
intersectsrelation. In fact, all 'traditional synonymy' relationships, cf. Berendsohn & al. (2000 [berendsohn_berlin_2003]), can be accommodated usingintersects. Also, citations of references in treatments are, in the context of TCS, best accommodated using theintersectsrelation.Examples
[TaxonConcept-intersects-example-1.ttl] [TaxonConcept-intersects-example-1.jsonld]
[TaxonConcept-intersects-example-2.ttl] [TaxonConcept-intersects-example-2.jsonld]
[TaxonConceptMapping-intersects-example-1.ttl] [TaxonConceptMapping-intersects-example-1.jsonld]
[TaxonConcept-intersects-example-3.ttl] [TaxonConcept-intersects-example-3.jsonld]