OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

lexidma message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Re: [lexidma] Working draft


Re gender etc, part of speech etc

I would recommend a key/value style representation of linguistic properties. This is the approach taken in all other models and would connect with other controlled vocabulary mechanisms (CLARIN, GOLD, LexInfo etc.)

Sounds promising. Can you explain in more detail what that would look like?


Re example sentences versus collocations, muti-wword units

I wouldnât say that it is just FGB. I think the practice is very common in bilingual dictionaries. For example, Collins Italian-English dictionary* has many full phrases. I was looking at the print version but the online one is similar:

https://www.collinsdictionary.com/dictionary/italian-english/questo

Right, OK.

Well, if the authors of a given dictionary donât want to make a distinction between examples on the one hand and collocations/multi-word combos on the other, then thatâs fine, they donât have to: they can treat them all as examples, and DMLex has a type for encoding examples. But for dictionary authors who do want to make such a dictinction, my suggestion would be that they make use of the subentrying mechanism in DMLex using SubentryRelation.


Re cross-referencing, arity and directionality of relations

What is the reason to care about symmetry and arity? It only seems useful if you are going to add some kind of reasoning or validation methodology, which would add many more complexities to the model.

It seemed like common sense to me to care about such things, I didnât even think about it â I guess this is where my object-oriented heritage is showing itself. :-) But yeah, good question. I guess itâs useful for encoding the facts we want to encode and for basic type safety, like everything else in DMLex. If we decide not to care about such things, then weâre going to end up allowing e.g. more than two participants in an antonymy relation, and thatâs nonsense, you canât have e.g. three words where each is an antonym of the other two. Or weâre going to end up having two words in a hypernym/hyponym relation and not knowing which is the hypernym and which is the hyponym.

M.



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]