pub fn rbc<I: Clone + Eq + Hash>(
results_a: &[(I, f32)],
results_b: &[(I, f32)],
) -> Vec<(I, f32)>Expand description
Rank-Biased Centroids (RBC) fusion.
Handles variable-length lists gracefully by using a geometric discount that depends on list length. More robust than RRF when lists have very different lengths.
Formula: score(d) = Σ (1 - p)^rank / (1 - p^N) where:
pis the persistence parameter (default 0.8, higher = more top-heavy)Nis the list lengthrankis 0-indexed
From Bailey et al. (2017). Better than RRF when lists have different lengths.