{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RP5TWVBDFBX43JG3XVYXVUJ5OK","short_pith_number":"pith:RP5TWVBD","schema_version":"1.0","canonical_sha256":"8bfb3b5423286fcda4dbbd717ad13d72b93f269ed672c532a56257e7382620c4","source":{"kind":"arxiv","id":"2503.20589","version":1},"attestation_state":"computed","paper":{"title":"What to Retrieve for Effective Retrieval-Augmented Code Generation? An Empirical Study and Beyond","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.SE","authors_text":"Juntao Chen, Mingwei Liu, Tianyue Jiang, Wenchao Gu, Xilin Liu, Xingzhe Li, Yanlin Wang, Yuchi Ma, Zibin Zheng","submitted_at":"2025-03-26T14:41:38Z","abstract_excerpt":"Repository-level code generation remains challenging due to complex code dependencies and the limitations of large language models (LLMs) in processing long contexts. While retrieval-augmented generation (RAG) frameworks are widely adopted, the effectiveness of different retrieved information sources-contextual code, APIs, and similar snippets-has not been rigorously analyzed. Through an empirical study on two benchmarks, we demonstrate that in-context code and potential API information significantly enhance LLM performance, whereas retrieved similar code often introduces noise, degrading resu"},"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":"2503.20589","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2025-03-26T14:41:38Z","cross_cats_sorted":[],"title_canon_sha256":"4f12d23836286dcbd6e91a3eed9e876c1768f18bf69bbec99bf24a5075bfe792","abstract_canon_sha256":"3857cdeb8341555f1f268ff536e5ecc3b4c43728537779713fe326624834b8b8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:39:40.891824Z","signature_b64":"eZhLhLTO8MaXpNmCS0COfIRwLn/4asb/zyR1xcrDc64FHcF5VnaD4eUuLTR4NejtmV+nzWvRWxMwXbizPZ5ABA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8bfb3b5423286fcda4dbbd717ad13d72b93f269ed672c532a56257e7382620c4","last_reissued_at":"2026-07-05T10:39:40.891302Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:39:40.891302Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What to Retrieve for Effective Retrieval-Augmented Code Generation? An Empirical Study and Beyond","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.SE","authors_text":"Juntao Chen, Mingwei Liu, Tianyue Jiang, Wenchao Gu, Xilin Liu, Xingzhe Li, Yanlin Wang, Yuchi Ma, Zibin Zheng","submitted_at":"2025-03-26T14:41:38Z","abstract_excerpt":"Repository-level code generation remains challenging due to complex code dependencies and the limitations of large language models (LLMs) in processing long contexts. While retrieval-augmented generation (RAG) frameworks are widely adopted, the effectiveness of different retrieved information sources-contextual code, APIs, and similar snippets-has not been rigorously analyzed. Through an empirical study on two benchmarks, we demonstrate that in-context code and potential API information significantly enhance LLM performance, whereas retrieved similar code often introduces noise, degrading resu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.20589","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/2503.20589/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":"2503.20589","created_at":"2026-07-05T10:39:40.891376+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.20589v1","created_at":"2026-07-05T10:39:40.891376+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.20589","created_at":"2026-07-05T10:39:40.891376+00:00"},{"alias_kind":"pith_short_12","alias_value":"RP5TWVBDFBX4","created_at":"2026-07-05T10:39:40.891376+00:00"},{"alias_kind":"pith_short_16","alias_value":"RP5TWVBDFBX43JG3","created_at":"2026-07-05T10:39:40.891376+00:00"},{"alias_kind":"pith_short_8","alias_value":"RP5TWVBD","created_at":"2026-07-05T10:39:40.891376+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08691","citing_title":"ProjAgent: Procedural Similarity Retrieval for Repository-Level Code Generation","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2606.08135","citing_title":"TICoder: A Repository-Level Code Generation Framework with Test-Driven Planning and Implementation-Aware Reuse","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31135","citing_title":"R+R: Reassessing Java Security API Misuse in Current LLMs: A Replication on JCA and JSSE APIs with External Security Knowledge","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17613","citing_title":"VeriCache: Turning Lossy KV Cache into Lossless LLM Inference","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14503","citing_title":"Not All RAGs Are Created Equal: A Component-Wise Empirical Study for Software Engineering Tasks","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05267","citing_title":"Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK","json":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK.json","graph_json":"https://pith.science/api/pith-number/RP5TWVBDFBX43JG3XVYXVUJ5OK/graph.json","events_json":"https://pith.science/api/pith-number/RP5TWVBDFBX43JG3XVYXVUJ5OK/events.json","paper":"https://pith.science/paper/RP5TWVBD"},"agent_actions":{"view_html":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK","download_json":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK.json","view_paper":"https://pith.science/paper/RP5TWVBD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.20589&json=true","fetch_graph":"https://pith.science/api/pith-number/RP5TWVBDFBX43JG3XVYXVUJ5OK/graph.json","fetch_events":"https://pith.science/api/pith-number/RP5TWVBDFBX43JG3XVYXVUJ5OK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK/action/storage_attestation","attest_author":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK/action/author_attestation","sign_citation":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK/action/citation_signature","submit_replication":"https://pith.science/pith/RP5TWVBDFBX43JG3XVYXVUJ5OK/action/replication_record"}},"created_at":"2026-07-05T10:39:40.891376+00:00","updated_at":"2026-07-05T10:39:40.891376+00:00"}