pub struct RecallConfig {
pub relevance_weight: f64,
pub importance_weight: f64,
pub temporal_weight: 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,
}Expand description
Configuration for the recall scoring pipeline. All fields have sensible defaults matching current behavior.
Fields§
§relevance_weight: f64Weight of RRF/fusion score. Default 0.70.
importance_weight: f64Weight of decay-adjusted salience. Default 0.20.
temporal_weight: f64Weight of pure recency. Default 0.10.
temporal_half_life_days: f64Days for temporal score to halve. Default 30.0.
decay_model: DecayModelDecay model to apply to salience. Default Exponential.
candidate_multiplier: u32Candidates 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: FusionStrategyStrategy used to fuse retrieval-source candidate lists. Default RRF k=60.
min_score: f64Minimum composite score to include in results. Default 0.0.
min_salience: f64Minimum raw salience to include in results. Default 0.0.
include_breakdown: boolInclude per-component score breakdowns in recall responses. Default false.
Implementations§
Source§impl RecallConfig
impl RecallConfig
Sourcepub fn validate(&self) -> Result<(), RuntimeError>
pub fn validate(&self) -> Result<(), RuntimeError>
Validate that the config is internally consistent.
Rejects:
- Negative weights
- All three weights summing to zero (no scoring signal)
- Non-positive temporal half-life
Trait Implementations§
Source§impl Clone for RecallConfig
impl Clone for RecallConfig
Source§fn clone(&self) -> RecallConfig
fn clone(&self) -> RecallConfig
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read more