Struct SearchExplainer 

pub struct SearchExplainer {
    pub rrf_k: f32,
    pub reranking_active: bool,
}

Generates SearchExplanation values for search results.

Fields

rrf_k: f32

RRF k parameter (should match SearchConfig::rrf_k).

reranking_active: bool

Whether a cross-encoder reranker was active during this search.

Implementations

impl SearchExplainer

pub fn new(rrf_k: f32, reranking_active: bool) -> Self

Create a new explainer.

pub fn explain_result( &self, memory_id: i64, rank: usize, bm25: f32, vector: f32, fuzzy: f32, recency: f32, importance: f32, rerank: Option<f32>, final_score: f32, ) -> SearchExplanation

Explain a single search result.

Parameters

• memory_id — ID of the memory being explained.
• rank — 1-based rank position in the result set.
• bm25 — BM25 keyword relevance score.
• vector — Vector/semantic similarity score.
• fuzzy — Fuzzy match score.
• recency — Recency boost applied.
• importance — Importance weight.
• rerank — Cross-encoder score (None if reranker inactive).
• final_score — Final combined score after fusion/reranking.

pub fn explain_batch( &self, results: Vec<(i64, f32, f32, f32, f32, f32, Option<f32>, f32)>, ) -> Vec<SearchExplanation>

Explain all results in a batch, assigning ranks 1..N in the order given.

Each tuple is (memory_id, bm25, vector, fuzzy, recency, importance, rerank, final_score).

pub fn generate_explanation( &self, rank: usize, scores: &ScoreBreakdown, signals: &[SignalContribution], ) -> String

Build the human-readable explanation string.

Trait Implementations

impl Default for SearchExplainer

fn default() -> Self

Returns the “default value” for a type.