evaluate

ucca.evaluation.evaluate(guessed, ref, converter=None, verbose=False, constructions={'implicit': <ucca.constructions.Construction object>, 'primary': <ucca.constructions.Construction object>, 'remote': <ucca.constructions.Construction object>}, units=False, fscore=True, errors=False, normalize=True, eval_type=None, ref_yield_tags=None, **kwargs)

Compare two passages and return requested diagnostics and scores, possibly printing them too. NOTE: since normalize=True by default, this method is destructive: it modifies the given passages before evaluation. :param guessed: Passage object to evaluate :param ref: reference Passage object to compare to :param converter: optional function to apply to passages before evaluation :param verbose: whether to print the results :param constructions: names of construction types to include in the evaluation :param units: whether to evaluate common units :param fscore: whether to compute precision, recall and f1 score :param errors: whether to print the mistakes :param normalize: flatten centers and move common functions to root before evaluation - modifies passages :param eval_type: specific evaluation type(s) to limit to :param ref_yield_tags: reference passage for fine-grained evaluation :return: Scores object