sqlite-vec/benchmarks-ann/bench-delete/README.md

bench-delete: Recall degradation after random deletion

Measures how KNN recall changes after deleting a random percentage of rows from different index types (flat, rescore, DiskANN).

Quick start

# Ensure dataset exists
make -C ../datasets/cohere1m

# Ensure extension is built
make -C ../.. loadable

# Quick smoke test
make smoke

# Full benchmark at 10k vectors
make bench-10k

Usage

python bench_delete.py --subset-size 10000 --delete-pct 10,25,50,75 \
  "flat:type=vec0-flat,variant=float" \
  "diskann-R72:type=diskann,R=72,L=128,quantizer=binary" \
  "rescore-bit:type=rescore,quantizer=bit,oversample=8"

What it measures

For each config and delete percentage:

Metric	Description
recall	KNN recall@k after deletion (ground truth recomputed over surviving rows)
delta	Recall change vs 0% baseline
query latency	Mean/median query time after deletion
db_size_mb	DB file size before VACUUM
vacuum_size_mb	DB file size after VACUUM (space reclaimed)
delete_time_s	Wall time for the DELETE operations

How it works

1. Build index with N vectors (one copy per config)
2. Measure recall at k=10 (pre-delete baseline)
3. For each delete %:
   • Copy the master DB
   • Delete a random selection of rows (deterministic seed)
   • Measure recall (ground truth recomputed over surviving rows only)
   • VACUUM and measure size savings
4. Print comparison table

Expected behavior

• Flat index: Recall should be 1.0 at all delete percentages (brute-force is always exact)
• Rescore: Recall should stay close to baseline (quantized scan + rescore is robust)
• DiskANN: Recall may degrade at high delete % due to graph fragmentation (dangling edges, broken connectivity)

Results DB

Results are stored in runs/<dataset>/<subset_size>/delete_results.db:

SELECT config_name, delete_pct, recall, vacuum_size_mb
FROM delete_runs
ORDER BY config_name, delete_pct;