Fusion blending model
Choosing between semantic and lexical results is a classic search challenge. Instead of relying only on a fixed heuristic such as RRF, our fusion model learns from historical dense and sparse outcomes. It can shift the balance between semantic breadth and lexical precision for each query before handing a stronger candidate set to the reranker.
Dense and sparse candidates are merged with rich features—scores, rank deltas, overlap, metadata—and trained against historical satisfaction signals.
The fusion model compresses the candidate set while recall, latency, and downstream token usage remain measurable.
Fusion outputs arrive calibrated, with score bands that plug straight into guardrails, dashboards, and feedback loops.
Learned fusion adjusts the balance of dense and sparse signals per query, handing the reranker a calibrated shortlist while keeping latency and token spend visible.
Intelligently blends dense and sparse candidates instead of relying on fixed heuristics.
Learns from historical dense and sparse outcomes so blending can reflect your domain.
Ranks by learned relevance, not raw similarity, giving the reranker a stronger shortlist.
Produces scores with clear thresholds so downstream systems know when to trust the set.
We replace Reciprocal Rank Fusion (RRF)—a fixed heuristic—with a learned model that produces a calibrated score. For a query q and candidate d, we learn:
Key inputs to the model:
Training optimizes pairwise or listwise losses (e.g., NDCG, MRR), and the resulting score is evaluated against relevance outcomes and downstream cost.