{"paper":{"title":"BBC: Improving Large-k Approximate Nearest Neighbor Search with a Bucket-based Result Collector","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A bucket-based collector organizes ANN candidates by distance to speed up large-k queries in quantization indexes by up to 3.8 times.","cross_cats":["cs.DS"],"primary_cat":"cs.DB","authors_text":"Bin Cui, Gao Cong, Jinwei Zhu, Kai Zeng, Ziqi Yin","submitted_at":"2026-04-02T12:22:38Z","abstract_excerpt":"Although Approximate Nearest Neighbor (ANN) search has been extensively studied, large-k ANN queries that aim to retrieve a large number of nearest neighbors remain underexplored, despite their numerous real-world applications. Existing ANN methods face significant performance degradation for such queries. In this work, we first investigate the reasons for the performance degradation of quantization-based ANN indexes: (1) the inefficiency of existing top-k collectors, which incurs significant overhead in candidate maintenance, and (2) the reduced pruning effectiveness of quantization methods, "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"BBC accelerates existing quantization-based ANN methods by up to 3.8x at recall@k = 0.95 for large-k ANN queries.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That organizing candidates into distance-based buckets reduces ranking costs and cache misses enough to offset any overhead, and that the two tailored re-ranking algorithms accelerate re-ranking across different quantization methods without harming accuracy.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BBC improves large-k ANN efficiency via bucketed candidate buffers and optimized re-ranking, delivering up to 3.8x speedup at recall@k=0.95.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A bucket-based collector organizes ANN candidates by distance to speed up large-k queries in quantization indexes by up to 3.8 times.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e13f9ba11c2bd7e3f8590534506c1625c4b92b3178cf6424f6073522005c47bd"},"source":{"id":"2604.01960","kind":"arxiv","version":3},"verdict":{"id":"561a2c56-bbd9-4e96-a68a-b8b001f54b24","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T20:43:39.806837Z","strongest_claim":"BBC accelerates existing quantization-based ANN methods by up to 3.8x at recall@k = 0.95 for large-k ANN queries.","one_line_summary":"BBC improves large-k ANN efficiency via bucketed candidate buffers and optimized re-ranking, delivering up to 3.8x speedup at recall@k=0.95.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That organizing candidates into distance-based buckets reduces ranking costs and cache misses enough to offset any overhead, and that the two tailored re-ranking algorithms accelerate re-ranking across different quantization methods without harming accuracy.","pith_extraction_headline":"A bucket-based collector organizes ANN candidates by distance to speed up large-k queries in quantization indexes by up to 3.8 times."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.01960/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"531dac0ff0d37ee7290368daa2677776a1c17856a41bb76dc41e2e4c47b471d7"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}