Struct SearchExplainer 

pub struct SearchExplainer { /* private fields */ }

Generates human-readable relevance explanations for semantic search results.

Implementations

impl SearchExplainer

pub fn new(config: ExplainerConfig) -> Self

Creates a new SearchExplainer with the provided configuration.

pub fn explain_result( &mut self, doc_id: &str, score: f64, contributions: Vec<ScoreContribution>, rank: usize, ctx: &QueryContext, ) -> ExplanationNode

Explains a single search result, returning a richly annotated ExplanationNode.

The method filters and sorts contributions according to the current ExplainerConfig, then generates explanation text.

pub fn explain_batch( &mut self, results: Vec<(String, f64, Vec<ScoreContribution>)>, ctx: &QueryContext, ) -> Vec<ExplanationNode>

Explains a batch of results, assigning ranks by order in results.

Each entry in results is (doc_id, score, contributions).

pub fn format_explanation(node: &ExplanationNode) -> String

Formats an ExplanationNode into a human-readable multi-line string.

pub fn top_contributions<'a>( node: &'a ExplanationNode, n: usize, ) -> Vec<&'a ScoreContribution>

Returns up to n contributions sorted by descending absolute weighted score.

pub fn score_breakdown( contributions: &[ScoreContribution], ) -> HashMap<String, f64>

Aggregates contributions by factor name, returning a map of factor → total weighted score.

pub fn filter_contributions( &self, contributions: Vec<ScoreContribution>, ) -> Vec<ScoreContribution>

Filters contributions according to min_contribution_weight and include_negative.

pub fn cosine_contribution( query_vec: &[f64], doc_vec: &[f64], weight: f64, ) -> ScoreContribution

Builds a ScoreContribution from the cosine similarity between two vectors.

Returns a zero-score contribution when either vector is empty or has zero norm.

pub fn term_frequency_contribution( term: &str, tf: f64, idf: f64, weight: f64, ) -> ScoreContribution

Builds a ScoreContribution from a TF-IDF component.

tf is term frequency, idf is inverse document frequency.

pub fn compare_explanations(a: &ExplanationNode, b: &ExplanationNode) -> String

Compares two ExplanationNodes and returns a textual diff explaining why the higher-ranked result outscored the lower-ranked one.

pub fn stats(&self) -> &ExplainerStats

Returns a reference to the current running statistics.