/partialED

Primary LanguagePython

Code for Event Detection with Partial Annotation

====

We develop a model that treats ED as a trigger localization problem (similar to a Machine Reading Comprehension model). The runing steps are:

  1. Run mrc_preprocessing.py for preprocessing.
  2. Run mrc_train.py for training with different ratios of labeled data.
  3. Run mrc_evaluation.py for evaluation a particular model.

We also provide code for the model SeqBERT for training, which has a similar training steps.

partial_labeled_data.zip contains the unlabelled event instances.