search

clamsproject / aapb-annotations

Repository to store manual annotation dataset developed for CLAMS-AAPB collaboration
3 stars 0 forks source link

adding raw annotation files to SR branch #68

Closed keighrim closed 8 months ago

keighrim commented 8 months ago

Starting a PR to merge raw files and batch files to SR readme branch (#66 ), so that all related information stay in a branch.

(once it's merged, I'd like to suggest to change the batch names from batch{A,B}-issue-number to issue-number-{suffix} to make them go together under OS file name sortings. For suffixes, how about using more meaningful ones than simple numbering, for example, tight/loose or fine/coarse ?

jarumihooi commented 8 months ago

So, one of the questions is this: Are batches meant to be reused for different things? If so, the dense and sparse as part of the name do not describe the batch itself but what it was subsequently used for in creating raw. If its not expected, then yes, better naming will make this much clearer. (Or if it doesn't matter if we don't care about how the names will be confusing in future usages for different projects.)

@keighrim

keighrim commented 8 months ago

Yes, they are supposed to be re-usable for different training and/or evaluation instances. dense/sparse also is a good suffix pair.

jarumihooi commented 8 months ago

Yes, they are supposed to be re-usable for different training and/or evaluation instances. dense/sparse also is a good suffix pair.

Hi, I want to confirm again: "dense/sparse" is a quality of how the batch was used for SR project specifically AFTER it was created. It does not describe the batch itself. Do we want to use this as part of the batch name?

keighrim commented 8 months ago

Good point. I completely misread the previous comment. You are right in that the batches are compiled without considering how (on which granularity) they would be annotated. Now I think simple numbering (A/B) aren't that bad choice for their names.

(I still prefer lower cases a/b since we have no upper case letters in other batch names so far and I'd like to propose a convention to use lower cases only for naming batches)

jarumihooi commented 8 months ago

Good point. I completely misread the previous comment. You are right in that the batches are compiled without considering how (on which granularity) they would be annotated. Now I think simple numbering (A/B) aren't that bad choice for their names.

(I still prefer lower cases a/b since we have no upper case letters in other batch names so far and I'd like to propose a convention to use lower cases only for naming batches)

Can make this a convention, will write it down for now.