Struct SemanticSummaryExtractor 

pub struct SemanticSummaryExtractor { /* private fields */ }

Extractive summariser driven by sentence embeddings.

Implementations

impl SemanticSummaryExtractor

pub fn new(config: ExtractorSummaryConfig) -> Self

Create a new extractor with the given configuration.

pub fn extract( &mut self, sentences: &[(String, Vec<f64>)], query_embedding: Option<&[f64]>, ) -> Result<ExtractionResult, String>

Extract the most representative sentences.

sentences is a slice of (text, embedding) pairs. If query_embedding is provided the initial scores are cosine similarities to the query; otherwise centrality scores are used.

pub fn cosine_similarity(a: &[f64], b: &[f64]) -> f64

Cosine similarity between two vectors.

Returns 0.0 when either vector has zero magnitude.

pub fn centrality_scores(embeddings: &[Vec<f64>]) -> Vec<f64>

Average cosine similarity of each embedding to all others (centrality).

pub fn coverage(selected: &[Vec<f64>], all: &[Vec<f64>]) -> f64

Coverage: for each sentence in all, compute the maximum cosine similarity to any sentence in selected, then return the average of those maxima.

Sentences that are themselves in selected are included in the average (their max similarity to a selected sentence is trivially 1.0).

pub fn stats(&self) -> SummaryExtractorStats

Return cumulative statistics.