BM25Similarity interface
Ranking function based on the Okapi BM25 similarity algorithm. BM25 is a TF-IDF-like algorithm that includes length normalization (controlled by the 'b' parameter) as well as term frequency saturation (controlled by the 'k1' parameter).
- Extends
Properties
b | This property controls how the length of a document affects the relevance score. By default, a value of 0.75 is used. A value of 0.0 means no length normalization is applied, while a value of 1.0 means the score is fully normalized by the length of the document. |
k1 | This property controls the scaling function between the term frequency of each matching terms and the final relevance score of a document-query pair. By default, a value of 1.2 is used. A value of 0.0 means the score does not scale with an increase in term frequency. |
odatatype | Polymorphic discriminator, which specifies the different types this object can be |
Property Details
b
This property controls how the length of a document affects the relevance score. By default, a value of 0.75 is used. A value of 0.0 means no length normalization is applied, while a value of 1.0 means the score is fully normalized by the length of the document.
b?: number
Property Value
number
k1
This property controls the scaling function between the term frequency of each matching terms and the final relevance score of a document-query pair. By default, a value of 1.2 is used. A value of 0.0 means the score does not scale with an increase in term frequency.
k1?: number
Property Value
number
odatatype
Polymorphic discriminator, which specifies the different types this object can be
odatatype: "#Microsoft.Azure.Search.BM25Similarity"
Property Value
"#Microsoft.Azure.Search.BM25Similarity"