{"paper":{"title":"From Schema to Signal: Retrieval-Augmented Modeling for Relational Data Analytics","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"RAM improves predictions on relational databases by augmenting schema graphs with semantic signals from tuple attributes via random walks and retrieval.","cross_cats":[],"primary_cat":"cs.DB","authors_text":"Beng Chin Ooi, Changshuo Liu, Lingze Zeng, Shaofeng Cai, Yuncheng Wu, Zhongle Xie","submitted_at":"2026-05-14T06:59:32Z","abstract_excerpt":"Relational data stored in RDBMS is foundational to many real-world applications across domains such as e-commerce, finance, and sociality. While deep neural networks (DNNs) have achieved strong performance on tabular data with a single table, extending these models to relational databases is challenging due to the normalized multi-table structure and complex inter-table relationships. Existing approaches often rely strictly on schema-defined graphs, which overlook implicit semantic signals embedded in tuple attributes and suffer from rigid connectivity.\n  In this work, we propose Retrieval-Aug"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments on five real-world relational databases demonstrate that RAM consistently outperforms existing baselines in diverse prediction tasks, establishing a state-of-the-art for relational data analytics.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That random-walk documents built from tuple attributes plus off-the-shelf IR relevance scores reliably surface semantically meaningful intra- and inter-table connections without introducing substantial noise or spurious correlations that degrade downstream learning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RAM augments relational graph models with attribute-semantic retrieval via random-walk documents and two contrastive augmentations (ATRA, ETRA) to achieve state-of-the-art results on five real-world databases.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RAM improves predictions on relational databases by augmenting schema graphs with semantic signals from tuple attributes via random walks and retrieval.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d758efb4d6e6b308881620edf0f1797f4d399f16febd10b34efbe46c61af4f8f"},"source":{"id":"2605.14464","kind":"arxiv","version":1},"verdict":{"id":"08997310-b0af-49a9-b379-37b748cded9e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:55:45.738270Z","strongest_claim":"Extensive experiments on five real-world relational databases demonstrate that RAM consistently outperforms existing baselines in diverse prediction tasks, establishing a state-of-the-art for relational data analytics.","one_line_summary":"RAM augments relational graph models with attribute-semantic retrieval via random-walk documents and two contrastive augmentations (ATRA, ETRA) to achieve state-of-the-art results on five real-world databases.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That random-walk documents built from tuple attributes plus off-the-shelf IR relevance scores reliably surface semantically meaningful intra- and inter-table connections without introducing substantial noise or spurious correlations that degrade downstream learning.","pith_extraction_headline":"RAM improves predictions on relational databases by augmenting schema graphs with semantic signals from tuple attributes via random walks and retrieval."},"references":{"count":47,"sample":[{"doi":"","year":2013,"title":"Event Recommendation Engine Challenge","work_id":"3aef96c0-2353-413a-b5f5-ad80e78ff8a5","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"Stack Exchange Data Dump","work_id":"c44434c6-ebd5-4bb5-9443-938dc6787ff5","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"Avito Context Ad Clicks","work_id":"12ad40d6-a9e4-4883-bac4-40ecd0175b24","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"AACT Clinical Trials.gove","work_id":"a4bcd505-4351-470a-827b-3543bf79665d","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Sercan Ö Arik and Tomas Pfister. 2021. Tabnet: Attentive interpretable tabular learning. InProceedings of the AAAI conference on artificial intelligence. AAAI Press, Palo Alto, California USA, 6679–66","work_id":"6041b0c6-67db-4da6-bee8-c1f5a21d95d8","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":47,"snapshot_sha256":"bff43303094362dd024b3cba9ae8174a2e94a114723ec4490369482b159f01d9","internal_anchors":5},"formal_canon":{"evidence_count":2,"snapshot_sha256":"fe36b68dbb96ed817b04503d77ff715799856265f2e01bfd44d1a097e0ef0b33"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}