Add option for dosdp generate to enable using synonyms for label generation

[matentzn]

What would be crazy awesome for NLP applications is this:

you have

UBERON:001 label "limb" 
exactSynonym "limb2"
relatedSynonym "limb3"

Would it be possible to get a setting that allows to generate like from likes?

Example: awesomeAnatomyPattern: R some X, NEW:X|UBERON:001

NEW:X label: "Awesome Limb"
NEW:X exactSynonym: "Awesome limb" (from label)
NEW:X exactSynonym: "Awesome limb2" (from syn)
NEW:X relatedSynonym: "Awesome limb3"

@cmungall double check if I capture cross-product correctly.

[cmungall]

Correct, but also the cross-product, e.g {awesome,brilliant} {limb,limb2,limb3}

You also need to define a small synonym scope algebra, e.g. ExE=E, ExL=E, RxE=R, BxB=B,Bx{E,R,N}=R,... and also something for types (we need a category for partial abbreviations).

You get a lot of noise, especially with chebi which doesn't respect obo conventions. I think the correct way to deal with this is to treat this as an upstream processing problem (e.g. filter or re-scope syns), and keep the logic and configuration in dosdp-tools simple

This is obviously also useful in LexParse (not yet implemented in dosdp tools, see the original obol approach)

[matentzn]

Lets see what Jim thinks, I am not that worried about where this code lives. I would like it to be in dosdp-tools as an optional option, but if I am the only one to do so, lets push it back until a more urgent use case calls for it.

[meghalithic]

+1 from me. It'd make it so that I wouldn't have to manually type in synonyms or track them down.

[aleixpuigb]

I would love to see this as well. It would be really useful for GWAS catalog. For example, at the moment, users that try searching for GWAS on 'level of protein X in blood', but using the synonym of protein X, can’t find it.