{"paper":{"title":"Derivation Prompting: A Logic-Based Method for Improving Retrieval-Augmented Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Derivation Prompting builds an interpretable logic tree from predefined rules to guide RAG generation and reduce unacceptable answers.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aiala Ros\\'a, Guillermo Moncecchi, Ignacio Sastre","submitted_at":"2026-05-13T19:20:16Z","abstract_excerpt":"The application of Large Language Models to Question Answering has shown great promise, but important challenges such as hallucinations and erroneous reasoning arise when using these models, particularly in knowledge-intensive, domain-specific tasks. To address these issues, we introduce Derivation Prompting, a novel prompting technique for the generation step of the Retrieval-Augmented Generation framework. Inspired by logic derivations, this method involves deriving conclusions from initial hypotheses through the systematic application of predefined rules. It constructs a derivation tree tha"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"It constructs a derivation tree that is interpretable and adds control over the generation process. We applied this method in a specific case study, significantly reducing unacceptable answers compared to traditional RAG and long-context window methods.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That predefined rules can be reliably encoded in prompts and followed by the LLM to produce valid, non-deviating derivation steps without introducing new errors or hallucinations in the tree construction itself.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Derivation Prompting constructs logic-based derivation trees in RAG generation to improve interpretability and reduce unacceptable answers compared to standard RAG or long-context methods in a case study.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Derivation Prompting builds an interpretable logic tree from predefined rules to guide RAG generation and reduce unacceptable answers.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"cf18f3aed523bf330c6f937c408ec782c5d96b1f8c2842a1a481c50d9955e5e9"},"source":{"id":"2605.14053","kind":"arxiv","version":1},"verdict":{"id":"a9051fd1-35bc-4374-b8fe-67bc92d6c440","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T05:26:46.825466Z","strongest_claim":"It constructs a derivation tree that is interpretable and adds control over the generation process. We applied this method in a specific case study, significantly reducing unacceptable answers compared to traditional RAG and long-context window methods.","one_line_summary":"Derivation Prompting constructs logic-based derivation trees in RAG generation to improve interpretability and reduce unacceptable answers compared to standard RAG or long-context methods in a case study.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That predefined rules can be reliably encoded in prompts and followed by the LLM to produce valid, non-deviating derivation steps without introducing new errors or hallucinations in the tree construction itself.","pith_extraction_headline":"Derivation Prompting builds an interpretable logic tree from predefined rules to guide RAG generation and reduce unacceptable answers."},"references":{"count":19,"sample":[{"doi":"","year":1901,"title":"In: Advances in Neural Infor- Derivation Prompting: A Logic-Based Method for Improving RAG 11 mation Processing Systems","work_id":"6076eda7-b5c4-49f0-90e3-f985b52c873b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Retrieval-Augmented Generation for Large Language Models: A Survey","work_id":"b80d2790-6cd9-4c87-b3c4-de404f99a80e","ref_index":2,"cited_arxiv_id":"2312.10997","is_internal_anchor":true},{"doi":"10.18653/v1/","year":2023,"title":"doi: 10.18653/v1/ 2024.findings-acl.586","work_id":"8d675bdd-79ca-48d6-9163-fc17ce0e8ece","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Izacard, G., Lewis, P., Lomeli, M., Hosseini, L., Petroni, F., Schick, T., Dwivedi- Yu, J., Joulin, A., Riedel, S., Grave, E.: Atlas: few-shot learning with retrieval augmented language models. J. Mac","work_id":"1c9a7fe4-c218-4280-b079-f66866a7cd2f","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1145/3571730","year":2023,"title":"Barret Zoph, Irwan Bello, Sameer Kumar, Nan Du, Yanping Huang, Jeff Dean, Noam Shazeer, and William Fedus","work_id":"aa2fb99f-beb8-463c-8cd4-c575e96e2512","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":19,"snapshot_sha256":"b07383cd63ea14b552e000749ade6f7a42c0849112f0e84ec875910d3450b5b8","internal_anchors":3},"formal_canon":{"evidence_count":2,"snapshot_sha256":"c4e81e499178d01d85ead281904d0725e33dbc985989410505468b0e066647ff"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}