We have train and test commands defined, which train the doc-ID model and evaluate it, respectively. We also need a classify command that applies the model to new data, but doesn't evaluate it. In order to reduce redundant code, the current test() function could perhaps use the classify() function to get a prediction, then use the gold label for comparison.
We have
trainandtestcommands defined, which train the doc-ID model and evaluate it, respectively. We also need aclassifycommand that applies the model to new data, but doesn't evaluate it. In order to reduce redundant code, the current test() function could perhaps use the classify() function to get a prediction, then use the gold label for comparison.