Datasets and results presented in the paper:"A semi-supervised approach to multi-label classification of NOTAMs using BERT"

Data:

labelled_dataset.xlsx: NOTAMs labelled with the multi-label system presented in "A semi-supervised approach to multi-label classification of NOTAMs using BERT"

unlabelled_dataset.xlsx: remaining non labelled NOTAMs

test_dataset.xlsx: NOTAMs used for testing the model (a subset of the labelled dataset).

test_dataset_zeroshot: the test_dataset including an additional column based on the outputs of LLaMa 3.1-7B when asked to assign labels to the NOTAMs. 

Csvs:

train.csv: labelled NOTAMs used for training in the needed format for the model.

test_predictions_bert_model2MERGED.csv: predictions of the BERT model on the test_dataset.

test_predictions_mixmatch_bert_model2MERGED (2).csv : predictions of the retrained BERT model on the test_dataset.