{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:Q7ZK5ZEXLUDWYMDULOBJW4Z6TM","short_pith_number":"pith:Q7ZK5ZEX","schema_version":"1.0","canonical_sha256":"87f2aee4975d076c30745b829b733e9b29a5b46d54d6989d7860e03f783bdff1","source":{"kind":"arxiv","id":"2505.18034","version":2},"attestation_state":"computed","paper":{"title":"Structured Thinking Matters: Improving LLMs Generalization in Causal Inference Tasks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Alonso Silva, Jo\\~ao Paulo Nogueira, Wentao Sun","submitted_at":"2025-05-23T15:37:40Z","abstract_excerpt":"Despite remarkable advances in the field, LLMs remain unreliable in distinguishing causation from correlation. Recent results from the Corr2Cause dataset benchmark reveal that state-of-the-art LLMs -- such as GPT-4 (F1 score: 29.08) -- only marginally outperform random baselines (Random Uniform, F1 score: 20.38), indicating limited capacity of generalization. To tackle this limitation, we propose a novel structured approach: rather than directly answering causal queries, we provide the model with the capability to structure its thinking by guiding the model to build a structured knowledge grap"},"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":"2505.18034","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-05-23T15:37:40Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"cab4fbbb3899e32c8f49ef927d74b33da7e6eaefb25007fb70969dfca357c8d2","abstract_canon_sha256":"f7c97090b2b7fd9f92c593c9f326f80da443f3faa96739f33408d4fae4eaced0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:10:09.637306Z","signature_b64":"0moUIV31svh8tPsyPaunVCe+LDYSSJfh5Xb4dA8Wxb9gP85drV73zccWkMucFIa5SaYXdvoUTYS/nraV6FnWDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"87f2aee4975d076c30745b829b733e9b29a5b46d54d6989d7860e03f783bdff1","last_reissued_at":"2026-07-05T11:10:09.636821Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:10:09.636821Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Structured Thinking Matters: Improving LLMs Generalization in Causal Inference Tasks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Alonso Silva, Jo\\~ao Paulo Nogueira, Wentao Sun","submitted_at":"2025-05-23T15:37:40Z","abstract_excerpt":"Despite remarkable advances in the field, LLMs remain unreliable in distinguishing causation from correlation. Recent results from the Corr2Cause dataset benchmark reveal that state-of-the-art LLMs -- such as GPT-4 (F1 score: 29.08) -- only marginally outperform random baselines (Random Uniform, F1 score: 20.38), indicating limited capacity of generalization. To tackle this limitation, we propose a novel structured approach: rather than directly answering causal queries, we provide the model with the capability to structure its thinking by guiding the model to build a structured knowledge grap"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.18034","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/2505.18034/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":"2505.18034","created_at":"2026-07-05T11:10:09.636876+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.18034v2","created_at":"2026-07-05T11:10:09.636876+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.18034","created_at":"2026-07-05T11:10:09.636876+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q7ZK5ZEXLUDW","created_at":"2026-07-05T11:10:09.636876+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q7ZK5ZEXLUDWYMDU","created_at":"2026-07-05T11:10:09.636876+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q7ZK5ZEX","created_at":"2026-07-05T11:10:09.636876+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.27567","citing_title":"Why LLMs Fail at Causal Discovery and How Interventional Agents Escape","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM","json":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM.json","graph_json":"https://pith.science/api/pith-number/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/graph.json","events_json":"https://pith.science/api/pith-number/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/events.json","paper":"https://pith.science/paper/Q7ZK5ZEX"},"agent_actions":{"view_html":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM","download_json":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM.json","view_paper":"https://pith.science/paper/Q7ZK5ZEX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.18034&json=true","fetch_graph":"https://pith.science/api/pith-number/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/graph.json","fetch_events":"https://pith.science/api/pith-number/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/action/storage_attestation","attest_author":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/action/author_attestation","sign_citation":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/action/citation_signature","submit_replication":"https://pith.science/pith/Q7ZK5ZEXLUDWYMDULOBJW4Z6TM/action/replication_record"}},"created_at":"2026-07-05T11:10:09.636876+00:00","updated_at":"2026-07-05T11:10:09.636876+00:00"}