{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:LD3LIH44WCA7ZHOD56DQYYV36O","short_pith_number":"pith:LD3LIH44","schema_version":"1.0","canonical_sha256":"58f6b41f9cb081fc9dc3ef870c62bbf3ad35e96f0c0921d65f22147b6ae14dcf","source":{"kind":"arxiv","id":"2202.08048","version":2},"attestation_state":"computed","paper":{"title":"Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.CL","authors_text":"Junjie Shan, Qi Zhang, Rui Zheng, Shihan Dou, Songyang Gao, Ting Wu, Xuanjing Huang, Yueming Wu","submitted_at":"2022-02-16T13:23:14Z","abstract_excerpt":"Natural language understanding (NLU) models tend to rely on spurious correlations (i.e., dataset bias) to achieve high performance on in-distribution datasets but poor performance on out-of-distribution ones. Most of the existing debiasing methods often identify and weaken these samples with biased features (i.e., superficial surface features that cause such spurious correlations). However, down-weighting these samples obstructs the model in learning from the non-biased parts of these samples. To tackle this challenge, in this paper, we propose to eliminate spurious correlations in a fine-grai"},"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":"2202.08048","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-02-16T13:23:14Z","cross_cats_sorted":["cs.AI","cs.IT","math.IT"],"title_canon_sha256":"eb399875b96e5e0b6c998a3c588f366d2e7db6eaa50dbc86a6aa39475a2795cc","abstract_canon_sha256":"30742052fcbbfc9f13a8a9e8c96e0e2000e7ea3f46e60608e8809b6e0e01e384"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:56:34.915677Z","signature_b64":"LqJfbT5FFSF6cjYHxeyYNcvD6+pDz7jXNyrqx70C8dR4HSevD3oeQdLQ81RGysKxNasNr47HAIqC2wBPTNgqCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"58f6b41f9cb081fc9dc3ef870c62bbf3ad35e96f0c0921d65f22147b6ae14dcf","last_reissued_at":"2026-07-05T04:56:34.915229Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:56:34.915229Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Decorrelate Irrelevant, Purify Relevant: Overcome Textual Spurious Correlations from a Feature Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.CL","authors_text":"Junjie Shan, Qi Zhang, Rui Zheng, Shihan Dou, Songyang Gao, Ting Wu, Xuanjing Huang, Yueming Wu","submitted_at":"2022-02-16T13:23:14Z","abstract_excerpt":"Natural language understanding (NLU) models tend to rely on spurious correlations (i.e., dataset bias) to achieve high performance on in-distribution datasets but poor performance on out-of-distribution ones. Most of the existing debiasing methods often identify and weaken these samples with biased features (i.e., superficial surface features that cause such spurious correlations). However, down-weighting these samples obstructs the model in learning from the non-biased parts of these samples. To tackle this challenge, in this paper, we propose to eliminate spurious correlations in a fine-grai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.08048","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/2202.08048/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":"2202.08048","created_at":"2026-07-05T04:56:34.915282+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.08048v2","created_at":"2026-07-05T04:56:34.915282+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.08048","created_at":"2026-07-05T04:56:34.915282+00:00"},{"alias_kind":"pith_short_12","alias_value":"LD3LIH44WCA7","created_at":"2026-07-05T04:56:34.915282+00:00"},{"alias_kind":"pith_short_16","alias_value":"LD3LIH44WCA7ZHOD","created_at":"2026-07-05T04:56:34.915282+00:00"},{"alias_kind":"pith_short_8","alias_value":"LD3LIH44","created_at":"2026-07-05T04:56:34.915282+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.05110","citing_title":"Rule Learning for Knowledge Graph Reasoning under Agnostic Distribution Shift","ref_index":54,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O","json":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O.json","graph_json":"https://pith.science/api/pith-number/LD3LIH44WCA7ZHOD56DQYYV36O/graph.json","events_json":"https://pith.science/api/pith-number/LD3LIH44WCA7ZHOD56DQYYV36O/events.json","paper":"https://pith.science/paper/LD3LIH44"},"agent_actions":{"view_html":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O","download_json":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O.json","view_paper":"https://pith.science/paper/LD3LIH44","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.08048&json=true","fetch_graph":"https://pith.science/api/pith-number/LD3LIH44WCA7ZHOD56DQYYV36O/graph.json","fetch_events":"https://pith.science/api/pith-number/LD3LIH44WCA7ZHOD56DQYYV36O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O/action/storage_attestation","attest_author":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O/action/author_attestation","sign_citation":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O/action/citation_signature","submit_replication":"https://pith.science/pith/LD3LIH44WCA7ZHOD56DQYYV36O/action/replication_record"}},"created_at":"2026-07-05T04:56:34.915282+00:00","updated_at":"2026-07-05T04:56:34.915282+00:00"}