train

Training Citron Models

Each component in Citron has a script which can be used to train and/or evaluate its associated model.

The scripts require data in Citron's Annotation Format. Citron provides pre-processing scripts to extract suitable data from the PARC 3.0 Corpus of Attribution Relations. Alternatively, you can create your own data using the Citron Annotator app.

• to train a model specify the --train-path parameter
• to evaluate a model specify the --test-path parameter

The exact match metrics require the predicted and correct spans to be an exact match. The overlap metrics are based on the number of predicted tokens that overlap with correct tokens, compared to the total number of predicted and correct tokens.

Usage

All scripts require the Citron project directory in the PYTHONPATH

$ export PYTHONPATH=$PYTHONPATH:/path/to/citron

All scripts share the following parameters:

    -h, --help    (Optional: show help message and exit)
    -v            (Optional: verbose mode)

Cue Classifier

Training a Cue Classifier requires a copy of VerbNet 3.3.

$ python3 cue_classifier_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)
    --verbnet-path    Path to VerbNet 3.3        (Optional: required to train)

Source Classifier

$ python3 source_classifier_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)

Source Resolver

$ python3 source_resolver_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)

Content Classifier

$ python3 content_classifier_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)

Content Resolver

$ python3 content_resolver_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)

Coreference Resolver

$ python3 coreference_resolver_builder.py
    --model-path      Path to model directory
    --train-path      Path to training data      (Optional: required to train)
    --test-path       Path to test data          (Optional: required to evaluate)

Copyright 2021 British Broadcasting Corporation.