{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:IVWWQRWXQYIG3ZOYBF6NEZ2CBP","short_pith_number":"pith:IVWWQRWX","schema_version":"1.0","canonical_sha256":"456d6846d786106de5d8097cd267420bf870495878ce0c6c6d642d5452baa78c","source":{"kind":"arxiv","id":"2209.14279","version":1},"attestation_state":"computed","paper":{"title":"Causal Proxy Models for Concept-Based Model Explanations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amir Zur, Atticus Geiger, Christopher Potts, Karel D'Oosterlinck, Zhengxuan Wu","submitted_at":"2022-09-28T17:45:07Z","abstract_excerpt":"Explainability methods for NLP systems encounter a version of the fundamental problem of causal inference: for a given ground-truth input text, we never truly observe the counterfactual texts necessary for isolating the causal effects of model representations on outputs. In response, many explainability methods make no use of counterfactual texts, assuming they will be unavailable. In this paper, we show that robust causal explainability methods can be created using approximate counterfactuals, which can be written by humans to approximate a specific counterfactual or simply sampled using meta"},"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":"2209.14279","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-09-28T17:45:07Z","cross_cats_sorted":[],"title_canon_sha256":"b21568483453674faa4e670d010227eb8d082cbcf4e2d2fcbb7b4efc655c6a3d","abstract_canon_sha256":"e908b0994915598624863b77745d868e4b12015e094a4039bce29d2d6609fe87"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:01:49.500034Z","signature_b64":"dpHJyh794eu8if1f3kaqnmSejWpNHjS/KnSq38D6wzFm8wl4vwVZeFF0F51hjSMG8hOgazqYbykJkY+Kn2k1AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"456d6846d786106de5d8097cd267420bf870495878ce0c6c6d642d5452baa78c","last_reissued_at":"2026-07-05T05:01:49.499623Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:01:49.499623Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Causal Proxy Models for Concept-Based Model Explanations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amir Zur, Atticus Geiger, Christopher Potts, Karel D'Oosterlinck, Zhengxuan Wu","submitted_at":"2022-09-28T17:45:07Z","abstract_excerpt":"Explainability methods for NLP systems encounter a version of the fundamental problem of causal inference: for a given ground-truth input text, we never truly observe the counterfactual texts necessary for isolating the causal effects of model representations on outputs. In response, many explainability methods make no use of counterfactual texts, assuming they will be unavailable. In this paper, we show that robust causal explainability methods can be created using approximate counterfactuals, which can be written by humans to approximate a specific counterfactual or simply sampled using meta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.14279","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/2209.14279/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":"2209.14279","created_at":"2026-07-05T05:01:49.499680+00:00"},{"alias_kind":"arxiv_version","alias_value":"2209.14279v1","created_at":"2026-07-05T05:01:49.499680+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.14279","created_at":"2026-07-05T05:01:49.499680+00:00"},{"alias_kind":"pith_short_12","alias_value":"IVWWQRWXQYIG","created_at":"2026-07-05T05:01:49.499680+00:00"},{"alias_kind":"pith_short_16","alias_value":"IVWWQRWXQYIG3ZOY","created_at":"2026-07-05T05:01:49.499680+00:00"},{"alias_kind":"pith_short_8","alias_value":"IVWWQRWX","created_at":"2026-07-05T05:01:49.499680+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.00555","citing_title":"On the Mechanistic Interpretability of Neural Networks for Causality in Bio-statistics","ref_index":70,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP","json":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP.json","graph_json":"https://pith.science/api/pith-number/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/graph.json","events_json":"https://pith.science/api/pith-number/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/events.json","paper":"https://pith.science/paper/IVWWQRWX"},"agent_actions":{"view_html":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP","download_json":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP.json","view_paper":"https://pith.science/paper/IVWWQRWX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2209.14279&json=true","fetch_graph":"https://pith.science/api/pith-number/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/graph.json","fetch_events":"https://pith.science/api/pith-number/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/action/storage_attestation","attest_author":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/action/author_attestation","sign_citation":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/action/citation_signature","submit_replication":"https://pith.science/pith/IVWWQRWXQYIG3ZOYBF6NEZ2CBP/action/replication_record"}},"created_at":"2026-07-05T05:01:49.499680+00:00","updated_at":"2026-07-05T05:01:49.499680+00:00"}