BM25

A probabilistic information retrieval model.

class src.features.bm25.BM25

Bases: object

A class to create BM25 features.

Methods: fit(corpus: pd.Series):

Train model

predict_proba(query: , document: ):

Return confidence score

bm25(word, document, k: int = 1, b: float = 0.75)

Compute weight

bm25(word, document, k: int = 1, b: float = 0.75)

Compute BM25 weight.

Parameters

• () (document) –

• () –

• k (int) –

• b (float) –

Returns

weight (float)

corpus = None

corpus_length = None

fit(corpus: pandas.core.series.Series)

Fits the model.

Parameters

corpus (pd.Series) –

Returns

none

l_avg = None

occurrences = {}

predict_proba(query, document)

Predict with confidence score.

Parameters

• () (document) –

• () –

Return type

score (float)