Could we add metaphor and metonym to the sense links for version1.4?

[fcbond]

I think we should have them go in two directions:

• metaphor (from target to source)

  • head (person) metaphor head (body part)
  • _metaphor is a relation between two senses, where the first sense is a metaphorical extension of the second sense_.

• has-metaphor (from source to target)

  • head (body part) has-metaphor head (person)
  • has-metaphor is a relation between two senses, where the first sense can be metaphorically extended to the second sense.

• metonym (from source to target)

  • chestnut (wood) metonym chestnut (tree)
  • metonym is a relation between two senses, where the first sense is a metonymic extension of the second sense.

• has-metonym (from target to source)

  • chestnut (tree) has-metonym chestnut (wood)

• metonym is a relation between two senses, where the first sense is can be metonymically extended to the second sense.

De Luca and Lönneker-Rodman (2008) proposed extMetaphorOf and extendedByMetaphor, but I think the simpler names I propose match the existing relations better.

Ernesto William De Luca and Birte Lönneker-Rodman. 2008. Integrating Metaphor Information into RDF/OWL EuroWordNet. In Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08), Marrakech, Morocco. European Language Resources Association (ELRA).

http://www.lrec-conf.org/proceedings/lrec2008/pdf/108_paper.pdf

@rowanhm and I willl have a paper about this in GWC 2025

[goodmami]

These sound like useful relations.

I think we should have them go in two directions

We've been linking things both ways for a while, but is it really necessary? Can we not just, e.g., traverse the metaphor relation in the opposite direction? I assume there must have been some practical purpose that I am now forgetting, but it seems like the reverse of many relations is entirely predictable given the original relation and an understanding of which relations have reverse links.

[fcbond]

Hmm.

So leave looking up the reverse to the tool/library?

I think PWN wanted to make a principled difference between those they considered reversible and those not (e.g. entailment). But (i) we could list this somewhere (ideally in the schema: reversible, bidirectional or nothing) and (ii) I want my tool to show the reverse of entailment anyway, ...

Some other wordnets (e.g. DanNet) prefer not to list reverse, and it would save space.

@jmccrae, @arademaker, @1313ou, @simongray, @piekvossen what do you think?

[jmccrae]

Our modelling always includes an inverse property (e.g., hypernym/hyponym) even when it is not necessary. OEWN is another resource that internally drops these inverse properties, but does include them in the XML export. It is well-documented in the GWA Docs which properties are inverse of each other.

I don't think this format requires all inverse properties to be included, and individual resources could choose not to include them if they wish.

Also @fcbond, I don't think your example of metonymy is actually a metonym. A metonym is 'the use of the name of one thing for that of another of which it is an attribute or with which it is associated', as such I understand at a relationship between two different lemmas not senses of the same lemma.

On the proposal overall however though, I think it is a great idea.👍

[fcbond]

Hi,

Our modelling always includes an inverse property (e.g., hypernym/hyponym) even when it is not necessary. OEWN is another resource that internally drops these inverse properties, but does include them in the XML export. It is well-documented in the GWA Docs which properties are inverse of each other. I don't think this format requires all inverse properties to be included, and individual resources could choose not to include them if they wish. Also @fcbond <https://github.com/fcbond>, I don't think your example of metonymy is actually a metonym. A metonym is 'the use of the name of one thing for that of another of which it is an attribute or with which it is associated' <https://www.merriam-webster.com/dictionary/metonymy>, as such I understand at a relationship between two different lemmas not senses of the same lemma.

I think my use is fairly standard, e.g., from wikipedia " Many cases of polysemy originate as metonyms: for example, "chicken" means the meat as well as the animal; "crown" for the object, as well as the institution."

…

On the proposal overall however though, I think it is a great idea.👍

-- Francis Bond <https://fcbond.github.io/>