def cosine_score(simple_index, query, document):
"""Oblicza cosine score."""
terms = set(simple_index.inverted_index.keys()).union(query)
doc_vector = [calc_tf_idf(simple_index, term, document) for term in terms]
qry_vector = []
for term in terms:
value = 0
if term in query:
value = calc_idf(simple_index, term) * query.count(term)
qry_vector.append(value)
num = sum(a*b for a, b in zip(doc_vector, qry_vector))
den = sqrt(sum(a*a for a in doc_vector))*sqrt(sum(a*a for a in qry_vector))
return num/den
def calc_tf_idf(simple_index, term, document):
"""Oblicza wartość TF-IDF."""
return calc_tf(simple_index, term, document) * calc_idf(simple_index, term)