{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:HHCDRKGXXLAENG7TQJW73PSEGF","short_pith_number":"pith:HHCDRKGX","schema_version":"1.0","canonical_sha256":"39c438a8d7bac0469bf3826dfdbe44317b878c78b66af0ab006965b573fc85c5","source":{"kind":"arxiv","id":"2012.15761","version":2},"attestation_state":"computed","paper":{"title":"Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bertie Vidgen, Douwe Kiela, Tristan Thrush, Zeerak Waseem","submitted_at":"2020-12-31T17:36:48Z","abstract_excerpt":"We present a human-and-model-in-the-loop process for dynamically generating datasets and training better performing and more robust hate detection models. We provide a new dataset of ~40,000 entries, generated and labelled by trained annotators over four rounds of dynamic data creation. It includes ~15,000 challenging perturbations and each hateful entry has fine-grained labels for the type and target of hate. Hateful entries make up 54% of the dataset, which is substantially higher than comparable datasets. We show that model performance is substantially improved using this approach. Models t"},"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":"2012.15761","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2020-12-31T17:36:48Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a9f7a8046e1fb889f3e542895a2d04fad264b6ea3544a653c8eda658d5b8b462","abstract_canon_sha256":"abd416482d598d47e322bc63763ab6ff165027e56203be60006a7c226c147255"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:45:49.270066Z","signature_b64":"qXqIFu2GCRuA8FFyMpdXzstIghwwdfegHwVu/DuRAapP8bJItQpVVfQckmc1MBJON0AgR/O+QE2oWxGv1t5mBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"39c438a8d7bac0469bf3826dfdbe44317b878c78b66af0ab006965b573fc85c5","last_reissued_at":"2026-07-05T02:45:49.269587Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:45:49.269587Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Bertie Vidgen, Douwe Kiela, Tristan Thrush, Zeerak Waseem","submitted_at":"2020-12-31T17:36:48Z","abstract_excerpt":"We present a human-and-model-in-the-loop process for dynamically generating datasets and training better performing and more robust hate detection models. We provide a new dataset of ~40,000 entries, generated and labelled by trained annotators over four rounds of dynamic data creation. It includes ~15,000 challenging perturbations and each hateful entry has fine-grained labels for the type and target of hate. Hateful entries make up 54% of the dataset, which is substantially higher than comparable datasets. We show that model performance is substantially improved using this approach. Models t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2012.15761","kind":"arxiv","version":2},"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/2012.15761/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":"2012.15761","created_at":"2026-07-05T02:45:49.269645+00:00"},{"alias_kind":"arxiv_version","alias_value":"2012.15761v2","created_at":"2026-07-05T02:45:49.269645+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2012.15761","created_at":"2026-07-05T02:45:49.269645+00:00"},{"alias_kind":"pith_short_12","alias_value":"HHCDRKGXXLAE","created_at":"2026-07-05T02:45:49.269645+00:00"},{"alias_kind":"pith_short_16","alias_value":"HHCDRKGXXLAENG7T","created_at":"2026-07-05T02:45:49.269645+00:00"},{"alias_kind":"pith_short_8","alias_value":"HHCDRKGX","created_at":"2026-07-05T02:45:49.269645+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20369","citing_title":"CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges","ref_index":289,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22435","citing_title":"Assisted Counterspeech Writing at the Crossroads of Hate Speech and Misinformation","ref_index":279,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF","json":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF.json","graph_json":"https://pith.science/api/pith-number/HHCDRKGXXLAENG7TQJW73PSEGF/graph.json","events_json":"https://pith.science/api/pith-number/HHCDRKGXXLAENG7TQJW73PSEGF/events.json","paper":"https://pith.science/paper/HHCDRKGX"},"agent_actions":{"view_html":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF","download_json":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF.json","view_paper":"https://pith.science/paper/HHCDRKGX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2012.15761&json=true","fetch_graph":"https://pith.science/api/pith-number/HHCDRKGXXLAENG7TQJW73PSEGF/graph.json","fetch_events":"https://pith.science/api/pith-number/HHCDRKGXXLAENG7TQJW73PSEGF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF/action/storage_attestation","attest_author":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF/action/author_attestation","sign_citation":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF/action/citation_signature","submit_replication":"https://pith.science/pith/HHCDRKGXXLAENG7TQJW73PSEGF/action/replication_record"}},"created_at":"2026-07-05T02:45:49.269645+00:00","updated_at":"2026-07-05T02:45:49.269645+00:00"}