rock

[arademaker]

PWN 3.0:

14696793-n rock, stone | (material consisting of the aggregate of minerals like those making up the Earth's crust; "that mountain is solid rock"; "stone is abundant in New England and there are many quarries")

09416076-n stone, rock | (a lump or mass of hard consolidated mineral matter; "he threw a rock at me")

could anyone explain the difference to me?

[arademaker]

The problem here is the inconsistencies of the hyponyms. See claystone that is a hyponym of 14696793-n rock but xenolith is hyponym of 09416076-n.

Note also that limestone (http://wnpt.sl.res.ibm.com/wn/synset?id=14936226-n) has the gloss a sedimentary rock consisting mainly of calcium that was deposited by the remains of marine animals) but it is not the hyponym of sedimentary_rock (http://wnpt.sl.res.ibm.com/wn/synset?id=14698000-n) as it should be.

[restinplace]

could anyone explain the difference to me?
First is the substance ("material..."), second is the object ("a lump or ...").
Asking seriously, is there a reason this is not clear from the defs?

[arademaker]

What was the decision? What will be changed? @restinplace, my second comment presents the problems. Definitions were clear but not consistent with the relations.

[jmccrae]

@arademaker I thought you were satisfied with @restinplace's explanation.

One synset refers to rock as a material, and the second a rock as an object. The hypernyms of 'claystone' and 'limestone' (as a material) and 'xenolith' (as a piece of rock) are fine to my understanding

[alexandretessarollo]

What about soapstone? 15044327-n Shouldn't it be under rock substance?

[jmccrae]

Currently 'soapstone' is a mineral (hence a material) but not a rock. I have no problem with adding this link as well.

Proposal: ewn-15068894-n is a hyponym of ewn-14720954-n

[arademaker]

The definitions of types of rocks are very confusing in PWN 3.0/3.1. Maybe xenolith was not a good example. But we may also be questioning clastic rock begin a type of the rock/object versus a sedimentary rock begin a type of rock/material.

To keep our focus on the problems already mentioned. Limestone need not be hyponym of rock (substance) if it is already hyponym of sedimentary rock.

[arademaker]

According to https://en.wikipedia.org/wiki/Soapstone, it is not a mineral but a type of metamorphic rock.

[jmccrae]

I asked a geology PhD: Apparently soapstone is a mineral but maybe also a rock. Essentially, a mineral is composed of a single chemical compound (which soapstone is), but it could be a rock because of how it is made. I wouldn't recommend changing anything here.

In English, it (in my opinion) is incorrect to say "a limestone", you would have to say "a lump of limestone". In contrast it is fine to say "a clastic rock" as well as a "lump of clastic rock", similarly with a "sedimentary rock". Probably both words should have both senses but this is a case of systematic polysemy and is a much bigger problem in the wordnet so we should probably postpone these changes.

The link between 'limestone' and 'rock' is superfluous as it is already covered by saying it is a 'sedimentary rock.' This could be removed.

[arademaker]

About the second paragraph:

CaCo3 cement, which may be fine grained (called micrite) or coarse grained (called sparite), holds the fossils (such as coccoliths and foraminifera) together to form a limestone.

From How is limestone formed?

Yes, systematic polysemy occurs a lot in the domain-specific texts. That is why such things are relevant to us. But I understand that this is not a priority for a resource designed to general-purpose language comprehension.

[jmccrae]

@arademaker What is the issue exactly?

[jmccrae]

For general discussion of systematic polysemy please use issue #243

[arademaker]

Well, I accepted the fact that this resource is not oriented to be a domain specific expansion. Given that, removing the redundant relation can close this issue.

But we are working on a branch of PWN to support language processing of texts from technical domain. In our case, a more systematic revision of the rocks and minerals hierarchy is needed. We are using ontologies and corpus as references.

[jmccrae]

@arademaker close this issue?

[arademaker]

@alexandretessarollo and I have a paper about our systematic review of all terms related to lithology. So yes, you can close this issue for now and later we can see how to sync our own PWN branch (https://github.com/own-pt/own-en) with this repo.