{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:U4LGYFAH64AOWJUESDHTOHQBLC","short_pith_number":"pith:U4LGYFAH","schema_version":"1.0","canonical_sha256":"a7166c1407f700eb268490cf371e01589aa0abd65b2f7ae1b2b493d3a12776fc","source":{"kind":"arxiv","id":"2606.01482","version":1},"attestation_state":"computed","paper":{"title":"Beyond Topical Similarity: Contrastive Evidence Retrieval with Interpretable Attention Alignment in RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ameeta Agrawal, Andr\\'e Freitas, Daniel Pedronette, Diego Alves, Francielle Vargas, Jo\\~ao Robiatti, Lucas Pascotti Valem, Maximilian Seeth, Sebasti\\'an Ferrada","submitted_at":"2026-05-31T22:54:44Z","abstract_excerpt":"Ensuring factuality and interpretability in RAG remains an open and urgent problem. We introduce Contrastive Evidence Rationale Attention (CERA), the first retrieval framework to employ subjectivity-based hard negative selection and inject an evidential inductive bias into contrastive learning through an auxiliary attention alignment loss. CERA fine-tunes a dense retriever using two training objectives: triplet-based contrastive learning and interpretable attention alignment, which supervises CLS-to-token attention using a part-of-speech-weighted masking distribution over human-annotated factu"},"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":"2606.01482","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-31T22:54:44Z","cross_cats_sorted":[],"title_canon_sha256":"d87a3f027775e93cfa1aaaf540639ceb6b16956b437b5f7f66dc27d28aae9494","abstract_canon_sha256":"a3981b41795051e7a0233d9cb81c43d613422a95977ec3419bed80073a340151"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:34.535905Z","signature_b64":"GjBCJfHdyg65rRiPVljbW83+jHjl5IAIHrt2qwxksCUidJlghEcS8E0zEuNxex0hT7YEScJaAcSLthUkZ+QVAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a7166c1407f700eb268490cf371e01589aa0abd65b2f7ae1b2b493d3a12776fc","last_reissued_at":"2026-06-02T02:04:34.535492Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:34.535492Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Topical Similarity: Contrastive Evidence Retrieval with Interpretable Attention Alignment in RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ameeta Agrawal, Andr\\'e Freitas, Daniel Pedronette, Diego Alves, Francielle Vargas, Jo\\~ao Robiatti, Lucas Pascotti Valem, Maximilian Seeth, Sebasti\\'an Ferrada","submitted_at":"2026-05-31T22:54:44Z","abstract_excerpt":"Ensuring factuality and interpretability in RAG remains an open and urgent problem. We introduce Contrastive Evidence Rationale Attention (CERA), the first retrieval framework to employ subjectivity-based hard negative selection and inject an evidential inductive bias into contrastive learning through an auxiliary attention alignment loss. CERA fine-tunes a dense retriever using two training objectives: triplet-based contrastive learning and interpretable attention alignment, which supervises CLS-to-token attention using a part-of-speech-weighted masking distribution over human-annotated factu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01482","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/2606.01482/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":"2606.01482","created_at":"2026-06-02T02:04:34.535553+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01482v1","created_at":"2026-06-02T02:04:34.535553+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01482","created_at":"2026-06-02T02:04:34.535553+00:00"},{"alias_kind":"pith_short_12","alias_value":"U4LGYFAH64AO","created_at":"2026-06-02T02:04:34.535553+00:00"},{"alias_kind":"pith_short_16","alias_value":"U4LGYFAH64AOWJUE","created_at":"2026-06-02T02:04:34.535553+00:00"},{"alias_kind":"pith_short_8","alias_value":"U4LGYFAH","created_at":"2026-06-02T02:04:34.535553+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC","json":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC.json","graph_json":"https://pith.science/api/pith-number/U4LGYFAH64AOWJUESDHTOHQBLC/graph.json","events_json":"https://pith.science/api/pith-number/U4LGYFAH64AOWJUESDHTOHQBLC/events.json","paper":"https://pith.science/paper/U4LGYFAH"},"agent_actions":{"view_html":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC","download_json":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC.json","view_paper":"https://pith.science/paper/U4LGYFAH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01482&json=true","fetch_graph":"https://pith.science/api/pith-number/U4LGYFAH64AOWJUESDHTOHQBLC/graph.json","fetch_events":"https://pith.science/api/pith-number/U4LGYFAH64AOWJUESDHTOHQBLC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC/action/storage_attestation","attest_author":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC/action/author_attestation","sign_citation":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC/action/citation_signature","submit_replication":"https://pith.science/pith/U4LGYFAH64AOWJUESDHTOHQBLC/action/replication_record"}},"created_at":"2026-06-02T02:04:34.535553+00:00","updated_at":"2026-06-02T02:04:34.535553+00:00"}