Documentation
¶
Index ¶
Constants ¶
View Source
const DefaultBM25B = 0.75
DefaultBM25B is the default document length normalization parameter.
View Source
const DefaultBM25K1 = 1.2
DefaultBM25K1 is the default term frequency saturation parameter.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type BM25Scorer ¶
BM25Scorer computes BM25 scores with configurable parameters.
func NewBM25Scorer ¶
func NewBM25Scorer(k1, b float64) *BM25Scorer
NewBM25Scorer creates a BM25Scorer with the given parameters. If k1 or b is <= 0, the default values (k1=1.2, b=0.75) are used.
func (*BM25Scorer) Score ¶
func (s *BM25Scorer) Score(queryTerms []string, docTermFreq map[string]int, docLen float64, avgDocLen float64, docCount int, docFreq map[string]int) float64
Score computes the BM25 score for a document given query terms. It computes IDF on the fly from docFreq and docCount.
Parameters:
- queryTerms: tokenized query terms
- docTermFreq: term frequency map for the document being scored
- docLen: the document's length (total term count)
- avgDocLen: average document length across the corpus
- docCount: total number of documents in the corpus (N)
- docFreq: number of documents containing each term (n(t))
Uses the standard BM25 formula:
sum over t in query of: IDF(t) * (tf * (k1+1)) / (tf + k1 * (1 - b + b * dl/avgdl))
with IDF(t) = log((N - n(t) + 0.5) / (n(t) + 0.5) + 1)
func (*BM25Scorer) ScoreWithIDF ¶
func (s *BM25Scorer) ScoreWithIDF(queryTerms []string, docTermFreq map[string]int, idf map[string]float64, docLen float64, avgDocLen float64) float64
ScoreWithIDF is a convenience method for callers that already have precomputed IDF values and know the document length directly.
Parameters:
- queryTerms: tokenized query terms
- docTermFreq: term frequency map for the document
- idf: precomputed IDF value for each term
- docLen: the document's length (total term count)
- avgDocLen: average document length across the corpus
Source Files
Click to show internal directories.
Click to hide internal directories.