Applying GNN-FiLM to dependency parsing

[LifeIsStrange]

Hi!
I am just a fan of state of the art, not an NLP researcher.

I found on NLP-progress this SOTA result on dependency parsing using GNN.
https://www.aclweb.org/anthology/P19-1237
Source: https://github.com/sebastianruder/NLP-progress/blob/master/english/dependency_parsing.md

I believe they use a GAT, and they claim to be the first to use a GNN for this fundamental NLP task.

I was thinking that, you made what seems to be like a major, task generic advance in GNNs, and that it would be interesting for you to make a collaboration with those researchers.

[LifeIsStrange]

They reach 97.3 in POS, 95.97 in UAS and 94.31 in LAS.
I dream of the day NLP tasks will have 99,99 % performance, and this collaboration could (I am maybe saying bullshit) make us closer to that goal: making NLP reliable.

[LifeIsStrange]

@mmjb

[mmjb]

This is an interesting idea, though I don't think that I'll have the time to include this dataset in the evaluation here anytime soon. I tried to have a look at the implementation listed in the paper, but the linked repository (https://github.com/AntNLP/gnn-dep-parsing) seems to not exist / be private. A first step may be to contact the authors to actually publish their code :-)