Struct RecallConfig 

pub struct RecallConfig {

    pub relevance_weight: f64,
    pub salience_weight: f64,
    pub temporal_weight: f64,
    pub reranker_weights: HashMap<String, f64>,
    pub temporal_half_life_days: f64,
    pub decay_model: DecayModel,
    pub candidate_multiplier: u32,
    pub candidate_limit: Option<u32>,
    pub fuse_strategy: FusionStrategy,
    pub min_score: f64,
    pub min_salience: f64,
    pub include_breakdown: bool,
    pub scoring: Option<ScoringConfig>,
    pub brain_profile: Option<BrainProfileHint>,
    pub fts_gather: RecallFtsGatherConfig,
    pub ann_overfetch_max_rounds: Option<usize>,
}

Configuration for the recall scoring pipeline. All fields have sensible defaults matching current behavior.

Fields

relevance_weight: f64

Weight of RRF/fusion score. Default 0.70.

salience_weight: f64

Weight of decay-adjusted salience. Default 0.20.

temporal_weight: f64

Weight of pure recency. Default 0.10.

reranker_weights: HashMap<String, f64>

Per-reranker weights, keyed by reranker name. Missing keys → 0.0 (disabled). v1 built-in names: “cross_encoder”, “salience”, “graph_proximity”.

temporal_half_life_days: f64

Days for temporal score to halve. Default 30.0.

decay_model: DecayModel

Decay model to apply to salience. Default Exponential.

candidate_multiplier: u32

Candidates per retrieval path before fusion = limit × this. Default 20.

candidate_limit: Option<u32>

Explicit max candidates per retrieval path before fusion. When None, candidate_multiplier keeps the legacy behavior.

fuse_strategy: FusionStrategy

Strategy used to fuse retrieval-source candidate lists. Default RRF k=60.

min_score: f64

Minimum composite score to include in results. Default 0.0.

min_salience: f64

Minimum raw salience to include in results. Default 0.0.

include_breakdown: bool

Include per-component score breakdowns in recall responses. Default false.

scoring: Option<ScoringConfig>

Optional archive scoring config; enables MMR, supersedes suppression, CJK routing, entity boost.

brain_profile: Option<BrainProfileHint>

Optional brain profile hint for post-recall score boosting.

fts_gather: RecallFtsGatherConfig

Controls the two-stage FTS gather path (default: disabled, existing behavior).

ann_overfetch_max_rounds: Option<usize>

Maximum rounds for the ANN namespace over-fetch retry loop.

Round 1 is the initial over-fetch; rounds 2–N double the fetch window until enough visible-namespace candidates are found or the corpus is exhausted. When None, falls back to the ANN_OVERFETCH_MAX_ROUNDS env var (default 3). Pass Some(1) to disable widening entirely.

Implementations

impl RecallConfig

pub fn validate(&self) -> Result<(), RuntimeError>

Validate config consistency: non-negative weights, positive weight sum, positive half-life.

pub fn try_from_value(v: Value) -> Result<Self, RuntimeError>

Deserialize from a JSON value and validate in one step.

Trait Implementations

impl Clone for RecallConfig

fn clone(&self) -> RecallConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RecallConfig

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

Formats the value using the given formatter.

impl Default for RecallConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for RecallConfig

where RecallConfig: Default,

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for RecallConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.