def characterize(bd, # this document
bj, # in this document collection
boost=param_dict(default=batch.BOOST),
k1=batch.K1,
b=param_dict(default=batch.B)):
words = set()
for bow in bd.values():
for w in bow.keys():
words.add(w)
pr = prf_result()
for w in words:
pr[w] = bm25f({w: 1}, bd, bj, boost, k1, b)
return pr
def test_k1(self):
self.assertAlmostEqual(
1.37142857142857 / (2.0 + 1.37142857142857) * -0.84729786038720 +
0.30769230769230 / (2.0 + 0.30769230769230) * 0.84729786038720,
bm25f(self.query, self.bd0, self.bj, k1=2.0))