{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:BVLW2ZGXBGHDSYXUB62PXVVH6A","short_pith_number":"pith:BVLW2ZGX","schema_version":"1.0","canonical_sha256":"0d576d64d7098e3962f40fb4fbd6a7f03a5471122507aec17e70ba21ccb228ed","source":{"kind":"arxiv","id":"2306.07607","version":1},"attestation_state":"computed","paper":{"title":"Practice with Graph-based ANN Algorithms on Sparse Data: Chi-square Two-tower model, HNSW, Sign Cauchy Projections","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.IR","authors_text":"Alice Wu, Chao Wang, Lijun Peng, Ping Li, Qi Xia, Weijie Zhao","submitted_at":"2023-06-13T08:05:30Z","abstract_excerpt":"Sparse data are common. The traditional ``handcrafted'' features are often sparse. Embedding vectors from trained models can also be very sparse, for example, embeddings trained via the ``ReLu'' activation function. In this paper, we report our exploration of efficient search in sparse data with graph-based ANN algorithms (e.g., HNSW, or SONG which is the GPU version of HNSW), which are popular in industrial practice, e.g., search and ads (advertising).\n  We experiment with the proprietary ads targeting application, as well as benchmark public datasets. For ads targeting, we train embeddings w"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2306.07607","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2023-06-13T08:05:30Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"560786f8530aa248c08efe809e47f1bb90ad18c7d8ee30c0dd97af5aa9153cf4","abstract_canon_sha256":"fe72ad9154a411d6d4b0418443abd60697d48afd38f994161c899560f3f169d4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:20:06.714644Z","signature_b64":"5O/tqv5W1UIs34fhva2mhYyTfxfWbvG6FHonvLuWdSBGLf1bHjUBwHsEzWHCjZyaUhgm1Wd/s3ZCFAiyOBS+DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0d576d64d7098e3962f40fb4fbd6a7f03a5471122507aec17e70ba21ccb228ed","last_reissued_at":"2026-07-05T06:20:06.714254Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:20:06.714254Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Practice with Graph-based ANN Algorithms on Sparse Data: Chi-square Two-tower model, HNSW, Sign Cauchy Projections","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.IR","authors_text":"Alice Wu, Chao Wang, Lijun Peng, Ping Li, Qi Xia, Weijie Zhao","submitted_at":"2023-06-13T08:05:30Z","abstract_excerpt":"Sparse data are common. The traditional ``handcrafted'' features are often sparse. Embedding vectors from trained models can also be very sparse, for example, embeddings trained via the ``ReLu'' activation function. In this paper, we report our exploration of efficient search in sparse data with graph-based ANN algorithms (e.g., HNSW, or SONG which is the GPU version of HNSW), which are popular in industrial practice, e.g., search and ads (advertising).\n  We experiment with the proprietary ads targeting application, as well as benchmark public datasets. For ads targeting, we train embeddings w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2306.07607","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2306.07607/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":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2306.07607","created_at":"2026-07-05T06:20:06.714324+00:00"},{"alias_kind":"arxiv_version","alias_value":"2306.07607v1","created_at":"2026-07-05T06:20:06.714324+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2306.07607","created_at":"2026-07-05T06:20:06.714324+00:00"},{"alias_kind":"pith_short_12","alias_value":"BVLW2ZGXBGHD","created_at":"2026-07-05T06:20:06.714324+00:00"},{"alias_kind":"pith_short_16","alias_value":"BVLW2ZGXBGHDSYXU","created_at":"2026-07-05T06:20:06.714324+00:00"},{"alias_kind":"pith_short_8","alias_value":"BVLW2ZGX","created_at":"2026-07-05T06:20:06.714324+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.14006","citing_title":"Experimental comparison of graph-based approximate nearest neighbor search algorithms on edge devices","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A","json":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A.json","graph_json":"https://pith.science/api/pith-number/BVLW2ZGXBGHDSYXUB62PXVVH6A/graph.json","events_json":"https://pith.science/api/pith-number/BVLW2ZGXBGHDSYXUB62PXVVH6A/events.json","paper":"https://pith.science/paper/BVLW2ZGX"},"agent_actions":{"view_html":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A","download_json":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A.json","view_paper":"https://pith.science/paper/BVLW2ZGX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2306.07607&json=true","fetch_graph":"https://pith.science/api/pith-number/BVLW2ZGXBGHDSYXUB62PXVVH6A/graph.json","fetch_events":"https://pith.science/api/pith-number/BVLW2ZGXBGHDSYXUB62PXVVH6A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A/action/storage_attestation","attest_author":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A/action/author_attestation","sign_citation":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A/action/citation_signature","submit_replication":"https://pith.science/pith/BVLW2ZGXBGHDSYXUB62PXVVH6A/action/replication_record"}},"created_at":"2026-07-05T06:20:06.714324+00:00","updated_at":"2026-07-05T06:20:06.714324+00:00"}