| name | hybrid-search |
| description | Use when combining vector (semantic) and full-text (lexical) search results into one ranking in a Go app with gosqlite — Reciprocal Rank Fusion via the fusion package. |
Hybrid search (semantic + lexical)
The fusion package merges two ranked result sets via Reciprocal Rank Fusion (Cormack 2009). Pure Go, no SQLite dependency.
import "gosqlite.org/fusion"
vecHits, _ := tbl.KNNSlice(ctx, queryVec, 50)
ftsHits, _ := idx.SearchSlice(ctx, fts.Term("brown fox"), fts.WithLimit(50))
vecKeys := make([]int64, len(vecHits))
for i, h := range vecHits { vecKeys[i] = h.Rowid }
ftsKeys := make([]int64, len(ftsHits))
for i, h := range ftsHits { ftsKeys[i] = h.Key }
top, _ := fusion.RRF([][]int64{vecKeys, ftsKeys}, fusion.WithLimit(20))
for _, r := range top { fmt.Println(r.Key, r.Score) }
fusion.RRF([][]K{...}, opts...) — fuse any number of ranked key lists.fusion.RRF2(a, b) — convenience for exactly two lists.- Keys must be the same comparable type across lists (the row identity, e.g.
int64 rowid or string).
Run the two searches with a generous limit (50–100) each, fuse, then take the top N. Prerequisites: the vector-search and full-text-search skills. Full reference: docs/guides/hybrid-search.md.
