{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:3CWB2SR7ELIQP4JSU7OC7TV7HB","short_pith_number":"pith:3CWB2SR7","schema_version":"1.0","canonical_sha256":"d8ac1d4a3f22d107f132a7dc2fcebf3860dce95ff13cec53a90e1af8ab8ab596","source":{"kind":"arxiv","id":"2305.15041","version":1},"attestation_state":"computed","paper":{"title":"Generating Faithful Synthetic Data with Large Language Models: A Case Study in Computational Social Science","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Akhil Arora, Ashton Anderson, Manoel Horta Ribeiro, Martin Josifoski, Robert West, Veniamin Veselovsky","submitted_at":"2023-05-24T11:27:59Z","abstract_excerpt":"Large Language Models (LLMs) have democratized synthetic data generation, which in turn has the potential to simplify and broaden a wide gamut of NLP tasks. Here, we tackle a pervasive problem in synthetic data generation: its generative distribution often differs from the distribution of real-world data researchers care about (in other words, it is unfaithful). In a case study on sarcasm detection, we study three strategies to increase the faithfulness of synthetic data: grounding, filtering, and taxonomy-based generation. We evaluate these strategies using the performance of classifiers trai"},"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":"2305.15041","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-05-24T11:27:59Z","cross_cats_sorted":[],"title_canon_sha256":"26b8ee02df91a05277c50d782f4c0c7128d3ab793223b88b3a84eb6b6023d244","abstract_canon_sha256":"935b68cdcbefb02969a26897de33811cc946351bf62579f572b13b59d6d6aa67"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:13:30.351365Z","signature_b64":"wFC51w/ElDPzTwNxwhT7pQQ1tvmTI6VNl+hsqMRjEEa710FQeyP7Pj/oxtUeJXlk8EnRqPhqqigP+UMT+18LBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d8ac1d4a3f22d107f132a7dc2fcebf3860dce95ff13cec53a90e1af8ab8ab596","last_reissued_at":"2026-07-05T06:13:30.350901Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:13:30.350901Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generating Faithful Synthetic Data with Large Language Models: A Case Study in Computational Social Science","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Akhil Arora, Ashton Anderson, Manoel Horta Ribeiro, Martin Josifoski, Robert West, Veniamin Veselovsky","submitted_at":"2023-05-24T11:27:59Z","abstract_excerpt":"Large Language Models (LLMs) have democratized synthetic data generation, which in turn has the potential to simplify and broaden a wide gamut of NLP tasks. Here, we tackle a pervasive problem in synthetic data generation: its generative distribution often differs from the distribution of real-world data researchers care about (in other words, it is unfaithful). In a case study on sarcasm detection, we study three strategies to increase the faithfulness of synthetic data: grounding, filtering, and taxonomy-based generation. We evaluate these strategies using the performance of classifiers trai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15041","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/2305.15041/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":"2305.15041","created_at":"2026-07-05T06:13:30.350957+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15041v1","created_at":"2026-07-05T06:13:30.350957+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15041","created_at":"2026-07-05T06:13:30.350957+00:00"},{"alias_kind":"pith_short_12","alias_value":"3CWB2SR7ELIQ","created_at":"2026-07-05T06:13:30.350957+00:00"},{"alias_kind":"pith_short_16","alias_value":"3CWB2SR7ELIQP4JS","created_at":"2026-07-05T06:13:30.350957+00:00"},{"alias_kind":"pith_short_8","alias_value":"3CWB2SR7","created_at":"2026-07-05T06:13:30.350957+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.27671","citing_title":"Evolving and Detecting Multi-Turn Deception using Geometric Signatures","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB","json":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB.json","graph_json":"https://pith.science/api/pith-number/3CWB2SR7ELIQP4JSU7OC7TV7HB/graph.json","events_json":"https://pith.science/api/pith-number/3CWB2SR7ELIQP4JSU7OC7TV7HB/events.json","paper":"https://pith.science/paper/3CWB2SR7"},"agent_actions":{"view_html":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB","download_json":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB.json","view_paper":"https://pith.science/paper/3CWB2SR7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15041&json=true","fetch_graph":"https://pith.science/api/pith-number/3CWB2SR7ELIQP4JSU7OC7TV7HB/graph.json","fetch_events":"https://pith.science/api/pith-number/3CWB2SR7ELIQP4JSU7OC7TV7HB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB/action/storage_attestation","attest_author":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB/action/author_attestation","sign_citation":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB/action/citation_signature","submit_replication":"https://pith.science/pith/3CWB2SR7ELIQP4JSU7OC7TV7HB/action/replication_record"}},"created_at":"2026-07-05T06:13:30.350957+00:00","updated_at":"2026-07-05T06:13:30.350957+00:00"}