{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:L7XYLX32GAUAAUOXZO2U7XGUAV","short_pith_number":"pith:L7XYLX32","schema_version":"1.0","canonical_sha256":"5fef85df7a30280051d7cbb54fdcd4055f556194cc5555f3c542d424b6ca4379","source":{"kind":"arxiv","id":"2505.06569","version":2},"attestation_state":"computed","paper":{"title":"MacRAG: Compress, Slice, and Scale-up for Multi-Scale Adaptive Context RAG","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Gyuwan Kim, Hyeonjung Kim, Jin Hyuk Lim, Kyungpyo Park, Kyuri Choi, Sungyoung Ji, William Yang Wang, Woosang Lim, Zekun Li","submitted_at":"2025-05-10T08:50:44Z","abstract_excerpt":"Long-context large language models (LC LLMs) combined with retrieval-augmented generation (RAG) hold strong potential for complex multi-hop and large-document tasks. However, existing RAG systems often suffer from imprecise retrieval, incomplete context coverage under constrained windows, and fragmented information from suboptimal context construction. We introduce Multi-scale Adaptive Context RAG (MacRAG), a hierarchical RAG framework that compresses and partitions documents into coarse-to-fine granularities, then adaptively merges relevant contexts through real-time chunk- and document-level"},"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":"2505.06569","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-10T08:50:44Z","cross_cats_sorted":["cs.AI","cs.IR","cs.LG"],"title_canon_sha256":"fa8e0acc4ea99dbd907b3ac8522a2c589de9f2c70c1afc46f2a05088f96dae63","abstract_canon_sha256":"b4230ecc62e1aef0dde6264d8540941eed65a576bd02f6754ab584f4083402cc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:31.939137Z","signature_b64":"uqpXoureAu3R3FX3+ZPs/TQchFoaRv8dq2znmL8cLXH0vwQTh90barExCbZA/WisHfI/VfN75KZGfeSH3vXIAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5fef85df7a30280051d7cbb54fdcd4055f556194cc5555f3c542d424b6ca4379","last_reissued_at":"2026-07-05T11:06:31.938668Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:31.938668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MacRAG: Compress, Slice, and Scale-up for Multi-Scale Adaptive Context RAG","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Gyuwan Kim, Hyeonjung Kim, Jin Hyuk Lim, Kyungpyo Park, Kyuri Choi, Sungyoung Ji, William Yang Wang, Woosang Lim, Zekun Li","submitted_at":"2025-05-10T08:50:44Z","abstract_excerpt":"Long-context large language models (LC LLMs) combined with retrieval-augmented generation (RAG) hold strong potential for complex multi-hop and large-document tasks. However, existing RAG systems often suffer from imprecise retrieval, incomplete context coverage under constrained windows, and fragmented information from suboptimal context construction. We introduce Multi-scale Adaptive Context RAG (MacRAG), a hierarchical RAG framework that compresses and partitions documents into coarse-to-fine granularities, then adaptively merges relevant contexts through real-time chunk- and document-level"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.06569","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/2505.06569/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":"2505.06569","created_at":"2026-07-05T11:06:31.938724+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.06569v2","created_at":"2026-07-05T11:06:31.938724+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.06569","created_at":"2026-07-05T11:06:31.938724+00:00"},{"alias_kind":"pith_short_12","alias_value":"L7XYLX32GAUA","created_at":"2026-07-05T11:06:31.938724+00:00"},{"alias_kind":"pith_short_16","alias_value":"L7XYLX32GAUAAUOX","created_at":"2026-07-05T11:06:31.938724+00:00"},{"alias_kind":"pith_short_8","alias_value":"L7XYLX32","created_at":"2026-07-05T11:06:31.938724+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.07458","citing_title":"CoinRAG: Contextualized Information Nugget KV Cache Reuse for Long-Context RAG","ref_index":34,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV","json":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV.json","graph_json":"https://pith.science/api/pith-number/L7XYLX32GAUAAUOXZO2U7XGUAV/graph.json","events_json":"https://pith.science/api/pith-number/L7XYLX32GAUAAUOXZO2U7XGUAV/events.json","paper":"https://pith.science/paper/L7XYLX32"},"agent_actions":{"view_html":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV","download_json":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV.json","view_paper":"https://pith.science/paper/L7XYLX32","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.06569&json=true","fetch_graph":"https://pith.science/api/pith-number/L7XYLX32GAUAAUOXZO2U7XGUAV/graph.json","fetch_events":"https://pith.science/api/pith-number/L7XYLX32GAUAAUOXZO2U7XGUAV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV/action/storage_attestation","attest_author":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV/action/author_attestation","sign_citation":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV/action/citation_signature","submit_replication":"https://pith.science/pith/L7XYLX32GAUAAUOXZO2U7XGUAV/action/replication_record"}},"created_at":"2026-07-05T11:06:31.938724+00:00","updated_at":"2026-07-05T11:06:31.938724+00:00"}