search

mickeysjm / TaxoClass

GNU General Public License v3.0
7 stars 0 forks source link

code release date? #1

Open diegoolano opened 3 years ago

diegoolano commented 3 years ago

Hello, I enjoyed this paper quite a bit and was wondering when you were planning on releasing the code?
Thank you for your time.

Decem-Y commented 3 years ago

Me, too!

NiceMartin commented 2 years ago

me 3.

YipingNUS commented 2 years ago

The first part of using a pretrained entailment model and some corpus stats should be straightforward for you to implement following the paper. The GNN part you can refer to the author's previous repo.

It's been a while since NAACL 2021 and I doubt the author is ever going to release the code. The next best thing is for you to implement it (partially) by yourself.

DavidLeexxxx commented 1 year ago

Actually, I have a question with the core class selection part if i try to implement this. In the NAACL paper, we need to choose L+2 sub-classes for each class at level L when we are selecting core class candidates. In fact, there are many circumstances which don't satisfy this requirement. Even the experiment dataset DBpedia mentioned in the paper can't guarantee this. So the selection strategy here seems quiet confusing.