Module summary_extractor 

Extractive summarization based on sentence embedding similarity.

SemanticSummaryExtractor selects representative sentences from a collection by scoring them against an optional query embedding (cosine similarity) or, when no query is provided, by centrality (average cosine similarity to every other sentence). A greedy selection loop with a configurable diversity penalty prevents redundant picks.

Structs

ExtractionResult

Result of an extraction run.

ExtractorScoredSentence

A sentence together with its embedding, score, and selection flag.

ExtractorSummaryConfig

Controls the behaviour of SemanticSummaryExtractor::extract.

SemanticSummaryExtractor

Extractive summariser driven by sentence embeddings.

SummaryExtractorStats

Simple stats counter.