{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:E2J3AKEAMGEDUPIRSILTAGEJSD","short_pith_number":"pith:E2J3AKEA","canonical_record":{"source":{"id":"2604.15591","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-04-17T00:09:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"23a60a146b20f1b8299ffab017b668508db696375ee51fb2bff2a1faec128224","abstract_canon_sha256":"b6c74cae47b4d0712fc640f7299f137ab8b044f2452aa027f152f0c4f3f34a65"},"schema_version":"1.0"},"canonical_sha256":"2693b0288061883a3d11921730188990cceac75908a57db0b41c502be8f98f37","source":{"kind":"arxiv","id":"2604.15591","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.15591","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"arxiv_version","alias_value":"2604.15591v2","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.15591","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_12","alias_value":"E2J3AKEAMGED","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_16","alias_value":"E2J3AKEAMGEDUPIR","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_8","alias_value":"E2J3AKEA","created_at":"2026-07-30T00:08:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:E2J3AKEAMGEDUPIRSILTAGEJSD","target":"record","payload":{"canonical_record":{"source":{"id":"2604.15591","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-04-17T00:09:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"23a60a146b20f1b8299ffab017b668508db696375ee51fb2bff2a1faec128224","abstract_canon_sha256":"b6c74cae47b4d0712fc640f7299f137ab8b044f2452aa027f152f0c4f3f34a65"},"schema_version":"1.0"},"canonical_sha256":"2693b0288061883a3d11921730188990cceac75908a57db0b41c502be8f98f37","receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2693b0288061883a3d11921730188990cceac75908a57db0b41c502be8f98f37","last_reissued_at":"2026-07-30T00:08:14.321755Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-30T00:08:14.321755Z"},"source_kind":"arxiv","source_id":"2604.15591","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-30T00:08:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tRHz1KfC2+7X0gL/TkKrabdeqySVU86N6oCcfc9dTobW3CeQW1dCPPMCoCI+6e/6kGyuuoB+j4ouPaKQbRMYAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T19:38:04.101771Z"},"content_sha256":"08979e3525f5272f7eade3158788bfcefe25494e391bb49e5a1816be75fee54f","schema_version":"1.0","event_id":"sha256:08979e3525f5272f7eade3158788bfcefe25494e391bb49e5a1816be75fee54f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:E2J3AKEAMGEDUPIRSILTAGEJSD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BioHiCL: Hierarchical Multi-Label Contrastive Learning for Biomedical Retrieval with MeSH Labels","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval.","cross_cats":["cs.AI"],"primary_cat":"cs.IR","authors_text":"Halil Kilicoglu, Lecheng Zheng, Mengfei Lan","submitted_at":"2026-04-17T00:09:01Z","abstract_excerpt":"Effective biomedical information retrieval requires modeling domain semantics and hierarchical relationships among biomedical texts. Existing biomedical generative retrievers build on coarse binary relevance signals, limiting their ability to capture semantic overlap. We propose BioHiCL (Biomedical Retrieval with Hierarchical Multi-Label Contrastive Learning), which leverages hierarchical MeSH annotations to provide structured supervision for multi-label contrastive learning. Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, senten"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, sentence similarity, and question answering tasks, while remaining computationally efficient for deployment.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That hierarchical MeSH annotations supply sufficiently structured and reliable supervision signals to outperform existing coarse binary relevance approaches in contrastive learning for retrieval.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BioHiCL applies hierarchical multi-label contrastive learning with MeSH annotations to improve biomedical retrieval, sentence similarity, and question answering using small efficient models.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"522441a89433cb2b20780030f4a861aee99a05cabf9820bf09dcb398f98c500e"},"source":{"id":"2604.15591","kind":"arxiv","version":2},"verdict":{"id":"05ae4405-b697-45b5-848a-d7b149afdac6","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T09:25:19.265461Z","strongest_claim":"Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, sentence similarity, and question answering tasks, while remaining computationally efficient for deployment.","one_line_summary":"BioHiCL applies hierarchical multi-label contrastive learning with MeSH annotations to improve biomedical retrieval, sentence similarity, and question answering using small efficient models.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That hierarchical MeSH annotations supply sufficiently structured and reliable supervision signals to outperform existing coarse binary relevance approaches in contrastive learning for retrieval.","pith_extraction_headline":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.15591/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"},"verdict_id":"05ae4405-b697-45b5-848a-d7b149afdac6"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-30T00:08:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0UmIcxMhA/uSCPk0FXNuPdICSjQ0EzreaDnYGYjg/HKbub/HEMdmn5bVC1IocwqWnMqq90q/sxBtgCpk7fJUAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-14T19:38:04.102426Z"},"content_sha256":"ad3fe36bcac4e223379f5d772b99310d99bae3fe317b62e3c5074a395a21b249","schema_version":"1.0","event_id":"sha256:ad3fe36bcac4e223379f5d772b99310d99bae3fe317b62e3c5074a395a21b249"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/bundle.json","state_url":"https://pith.science/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-14T19:38:04Z","links":{"resolver":"https://pith.science/pith/E2J3AKEAMGEDUPIRSILTAGEJSD","bundle":"https://pith.science/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/bundle.json","state":"https://pith.science/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E2J3AKEAMGEDUPIRSILTAGEJSD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:E2J3AKEAMGEDUPIRSILTAGEJSD","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"b6c74cae47b4d0712fc640f7299f137ab8b044f2452aa027f152f0c4f3f34a65","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-04-17T00:09:01Z","title_canon_sha256":"23a60a146b20f1b8299ffab017b668508db696375ee51fb2bff2a1faec128224"},"schema_version":"1.0","source":{"id":"2604.15591","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.15591","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"arxiv_version","alias_value":"2604.15591v2","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.15591","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_12","alias_value":"E2J3AKEAMGED","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_16","alias_value":"E2J3AKEAMGEDUPIR","created_at":"2026-07-30T00:08:14Z"},{"alias_kind":"pith_short_8","alias_value":"E2J3AKEA","created_at":"2026-07-30T00:08:14Z"}],"graph_snapshots":[{"event_id":"sha256:ad3fe36bcac4e223379f5d772b99310d99bae3fe317b62e3c5074a395a21b249","target":"graph","created_at":"2026-07-30T00:08:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, sentence similarity, and question answering tasks, while remaining computationally efficient for deployment."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That hierarchical MeSH annotations supply sufficiently structured and reliable supervision signals to outperform existing coarse binary relevance approaches in contrastive learning for retrieval."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"BioHiCL applies hierarchical multi-label contrastive learning with MeSH annotations to improve biomedical retrieval, sentence similarity, and question answering using small efficient models."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval."}],"snapshot_sha256":"522441a89433cb2b20780030f4a861aee99a05cabf9820bf09dcb398f98c500e"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.15591/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Effective biomedical information retrieval requires modeling domain semantics and hierarchical relationships among biomedical texts. Existing biomedical generative retrievers build on coarse binary relevance signals, limiting their ability to capture semantic overlap. We propose BioHiCL (Biomedical Retrieval with Hierarchical Multi-Label Contrastive Learning), which leverages hierarchical MeSH annotations to provide structured supervision for multi-label contrastive learning. Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, senten","authors_text":"Halil Kilicoglu, Lecheng Zheng, Mengfei Lan","cross_cats":["cs.AI"],"headline":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-04-17T00:09:01Z","title":"BioHiCL: Hierarchical Multi-Label Contrastive Learning for Biomedical Retrieval with MeSH Labels"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.15591","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T09:25:19.265461Z","id":"05ae4405-b697-45b5-848a-d7b149afdac6","model_set":{"reader":"grok-4.3"},"one_line_summary":"BioHiCL applies hierarchical multi-label contrastive learning with MeSH annotations to improve biomedical retrieval, sentence similarity, and question answering using small efficient models.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Hierarchical MeSH labels supply structured supervision for multi-label contrastive learning that improves biomedical retrieval.","strongest_claim":"Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, sentence similarity, and question answering tasks, while remaining computationally efficient for deployment.","weakest_assumption":"That hierarchical MeSH annotations supply sufficiently structured and reliable supervision signals to outperform existing coarse binary relevance approaches in contrastive learning for retrieval."}},"verdict_id":"05ae4405-b697-45b5-848a-d7b149afdac6"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:08979e3525f5272f7eade3158788bfcefe25494e391bb49e5a1816be75fee54f","target":"record","created_at":"2026-07-30T00:08:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"b6c74cae47b4d0712fc640f7299f137ab8b044f2452aa027f152f0c4f3f34a65","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2026-04-17T00:09:01Z","title_canon_sha256":"23a60a146b20f1b8299ffab017b668508db696375ee51fb2bff2a1faec128224"},"schema_version":"1.0","source":{"id":"2604.15591","kind":"arxiv","version":2}},"canonical_sha256":"2693b0288061883a3d11921730188990cceac75908a57db0b41c502be8f98f37","receipt":{"builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2693b0288061883a3d11921730188990cceac75908a57db0b41c502be8f98f37","first_computed_at":"2026-07-30T00:08:14.321755Z","kind":"pith_receipt","last_reissued_at":"2026-07-30T00:08:14.321755Z","receipt_version":"0.3","signature_status":"unsigned_v0"},"source_id":"2604.15591","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:08979e3525f5272f7eade3158788bfcefe25494e391bb49e5a1816be75fee54f","sha256:ad3fe36bcac4e223379f5d772b99310d99bae3fe317b62e3c5074a395a21b249"],"state_sha256":"1949f7ba612cae83207f60568c561c4fa33ea8695bfdd343f6fca365a5e2c59f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"McuWU0dE1Qa9S6Ts9OdI5nwk/R1PK8jNM2ZthpZZXUefUtQ2LZsVj3LQHmnGZUDd5aAeRXQ9R33Y6NDIKlHCDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-14T19:38:04.108305Z","bundle_sha256":"9e420037a7eceac2acf6d3262480ea94735f2c5d57820c9bcb70b75631ce7252"}}