search

mmxgn / sprl-spacy

Implementation of Spatial Role Labeling using the Spacy NLP framework.
19 stars 0 forks source link

Are there any detailed explanation for the model specification? #4

Open jellyheadandrew opened 2 years ago

jellyheadandrew commented 2 years ago

Hello! Thanks for your nice work on spatial role labeling. I'm using it for my task and it is very useful.

I'd like to reference your work, but I found no model specification in this repository.

Are there (or will there be) any detailed explanation for your work?? For example, I'd like to know what extra information about your model, e.g. model architecture (e.g. transformer), how you trained the models, etc.

Thanks again!

mmxgn commented 2 years ago

Hi,

Thank you for your interest.

As I say on the readme the model features are given in. The entity tagging has been done with SpaCy's engine and the relation extraction using an SVM. You could unpickle the models and observe them in python for details.

So you should cite SpaCy, and also:

Nichols, Eric, and Fadi Botros. 
"SpRL-CWW: Spatial relation classification with independent multi-class models." 
Proceedings of the 9th International Workshop on Semantic Evaluation.

If you want to cite this work cite the above paper and mention that you are using the specific implementation found in this repository

mmxgn commented 2 years ago

Alternatively, if you could wait, I could try when I have time this week to add a zenodo doi and a citation file as well as a bit of description.

jellyheadandrew commented 2 years ago

Oh, I could wait. Thnks