{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:VOCXG5VK4AXFDDPDSUYWKNA327","short_pith_number":"pith:VOCXG5VK","schema_version":"1.0","canonical_sha256":"ab857376aae02e518de3953165341bd7e7142458d5b2ddb477de7bd788baed7e","source":{"kind":"arxiv","id":"2210.15221","version":1},"attestation_state":"computed","paper":{"title":"TASA: Deceiving Question Answering Models by Twin Answer Sentences Attack","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dacheng Tao, Dianqi Li, Jun Gao, Meng Fang, Tianyi Zhou, Yibing Zhan, Yu Cao","submitted_at":"2022-10-27T07:16:30Z","abstract_excerpt":"We present Twin Answer Sentences Attack (TASA), an adversarial attack method for question answering (QA) models that produces fluent and grammatical adversarial contexts while maintaining gold answers. Despite phenomenal progress on general adversarial attacks, few works have investigated the vulnerability and attack specifically for QA models. In this work, we first explore the biases in the existing models and discover that they mainly rely on keyword matching between the question and context, and ignore the relevant contextual relations for answer prediction. Based on two biases above, TASA"},"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":"2210.15221","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-10-27T07:16:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d83c1c3bcf5e1c9e076e8a0f4102a4faf63376aa209a8309d951e88c06201bac","abstract_canon_sha256":"0f3309fb9f9f3506ae35686a8394a89701b05afd99287f6e88efddb9ae2bdc5b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:11:03.790553Z","signature_b64":"f5OwtPHGxM+8+Rsj0Fzz8MZQX9WubsSSg2rZ9herujmaTZtSZdPHw24M1WW6YUi7BUF6a6BC9eXSYCvVU8iYAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ab857376aae02e518de3953165341bd7e7142458d5b2ddb477de7bd788baed7e","last_reissued_at":"2026-07-05T05:11:03.790013Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:11:03.790013Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TASA: Deceiving Question Answering Models by Twin Answer Sentences Attack","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dacheng Tao, Dianqi Li, Jun Gao, Meng Fang, Tianyi Zhou, Yibing Zhan, Yu Cao","submitted_at":"2022-10-27T07:16:30Z","abstract_excerpt":"We present Twin Answer Sentences Attack (TASA), an adversarial attack method for question answering (QA) models that produces fluent and grammatical adversarial contexts while maintaining gold answers. Despite phenomenal progress on general adversarial attacks, few works have investigated the vulnerability and attack specifically for QA models. In this work, we first explore the biases in the existing models and discover that they mainly rely on keyword matching between the question and context, and ignore the relevant contextual relations for answer prediction. Based on two biases above, TASA"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.15221","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/2210.15221/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":"2210.15221","created_at":"2026-07-05T05:11:03.790074+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.15221v1","created_at":"2026-07-05T05:11:03.790074+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.15221","created_at":"2026-07-05T05:11:03.790074+00:00"},{"alias_kind":"pith_short_12","alias_value":"VOCXG5VK4AXF","created_at":"2026-07-05T05:11:03.790074+00:00"},{"alias_kind":"pith_short_16","alias_value":"VOCXG5VK4AXFDDPD","created_at":"2026-07-05T05:11:03.790074+00:00"},{"alias_kind":"pith_short_8","alias_value":"VOCXG5VK","created_at":"2026-07-05T05:11:03.790074+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2406.09250","citing_title":"MirrorCheck: Efficient Adversarial Defense for Vision-Language Models","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327","json":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327.json","graph_json":"https://pith.science/api/pith-number/VOCXG5VK4AXFDDPDSUYWKNA327/graph.json","events_json":"https://pith.science/api/pith-number/VOCXG5VK4AXFDDPDSUYWKNA327/events.json","paper":"https://pith.science/paper/VOCXG5VK"},"agent_actions":{"view_html":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327","download_json":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327.json","view_paper":"https://pith.science/paper/VOCXG5VK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.15221&json=true","fetch_graph":"https://pith.science/api/pith-number/VOCXG5VK4AXFDDPDSUYWKNA327/graph.json","fetch_events":"https://pith.science/api/pith-number/VOCXG5VK4AXFDDPDSUYWKNA327/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327/action/storage_attestation","attest_author":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327/action/author_attestation","sign_citation":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327/action/citation_signature","submit_replication":"https://pith.science/pith/VOCXG5VK4AXFDDPDSUYWKNA327/action/replication_record"}},"created_at":"2026-07-05T05:11:03.790074+00:00","updated_at":"2026-07-05T05:11:03.790074+00:00"}