Struct RerankFeatures 

pub struct RerankFeatures {
    pub relevance: f64,
    pub salience: f64,
    pub temporal: f64,
    pub text_match: bool,
    pub vector_match: bool,
}

Input features available per recall candidate for weighted reranking.

Each field is a normalized scalar in roughly [0, 1] (though temporal and salience can technically exceed 1.0 only if the raw salience > 1.0 or age < 0, neither of which the runtime produces in practice).

Supported feature names (keys in reranker_weights):

• "relevance" — fused retrieval score (RRF/weighted fusion output)
• "salience" — note salience after decay (salience * exp(-k * age))
• "temporal" — recency score (exp(-ln2/half_life * age_days))
• "text_match" — 1.0 when candidate appeared in FTS text results, else 0.0
• "vector_match" — 1.0 when candidate appeared in vector results, else 0.0

Fields

relevance: f64

Fused retrieval score from RRF or weighted fusion.

salience: f64

Decay-adjusted salience value (raw salience × decay factor).

temporal: f64

Half-life–decay recency score independent of per-note decay_factor.

text_match: bool

True when candidate appeared in FTS text search results.

vector_match: bool

True when candidate appeared in vector search results.

Trait Implementations

impl Clone for RerankFeatures

fn clone(&self) -> RerankFeatures

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for RerankFeatures

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.