def compute_ngram_scores(candidate: Sequence[Any],
reference: Sequence[Any],
n: int = 4) -> Score:
"""
Compute the score based on ngram co-occurence of sequences of items
Args:
candidate: candidate sequence of items
reference: reference sequence of items
n: ngram order
Returns:
The score containing the number of ngram co-occurences
.. versionadded:: 0.4.5
"""
# ngrams of the candidate
candidate_counter = ngrams(candidate, n)
# ngrams of the references
reference_counter = ngrams(reference, n)
# ngram co-occurences in the candidate and the references
match_counters = candidate_counter & reference_counter
# the score is defined using Fraction
return Score(
match=sum(match_counters.values()),
candidate=sum(candidate_counter.values()),
reference=sum(reference_counter.values()),
)
def test_ngrams(sequence, n, expected_keys, expected_values):
ngrams_counter = ngrams(sequence=sequence, n=n)
assert list(ngrams_counter.values()) == expected_values
assert list(ngrams_counter.keys()) == expected_keys
