{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:I345H75Q2A5BN6ZUDPV3VWAOUC","short_pith_number":"pith:I345H75Q","schema_version":"1.0","canonical_sha256":"46f9d3ffb0d03a16fb341bebbad80ea0a47b4062b97061407c38783053958c4b","source":{"kind":"arxiv","id":"1709.00103","version":7},"attestation_state":"computed","paper":{"title":"Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Seq2SQL translates natural language questions into SQL queries by combining structured generation with reinforcement learning rewards from database executions, reaching 59.4 percent execution accuracy.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Caiming Xiong, Richard Socher, Victor Zhong","submitted_at":"2017-08-31T23:12:15Z","abstract_excerpt":"A significant amount of the world's knowledge is stored in relational databases. However, the ability for users to retrieve facts from a database is limited due to a lack of understanding of query languages such as SQL. We propose Seq2SQL, a deep neural network for translating natural language questions to corresponding SQL queries. Our model leverages the structure of SQL queries to significantly reduce the output space of generated queries. Moreover, we use rewards from in-the-loop query execution over the database to learn a policy to generate unordered parts of the query, which we show are"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"1709.00103","kind":"arxiv","version":7},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2017-08-31T23:12:15Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1a5ccd4fc7bdb0c7f10e9cb3ff6460df4ab865e742bccf6b281a0bf23dd02796","abstract_canon_sha256":"634462eb8ea9f932ce45f70ef44f9c4918af5f660aadc44e312a5c09d7767d75"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-04T22:19:56.634397Z","signature_b64":"A1h6dUWEZwUK+3Q4qy/lSzm1OlsxkdRwQz2dlpx+Z0i0/enfsSnNibXSuu2P8aZgOYqMks5USjbxO3YhnH1MBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"46f9d3ffb0d03a16fb341bebbad80ea0a47b4062b97061407c38783053958c4b","last_reissued_at":"2026-07-04T22:19:56.633883Z","signature_status":"signed_v1","first_computed_at":"2026-07-04T22:19:56.633883Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Seq2SQL translates natural language questions into SQL queries by combining structured generation with reinforcement learning rewards from database executions, reaching 59.4 percent execution accuracy.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Caiming Xiong, Richard Socher, Victor Zhong","submitted_at":"2017-08-31T23:12:15Z","abstract_excerpt":"A significant amount of the world's knowledge is stored in relational databases. However, the ability for users to retrieve facts from a database is limited due to a lack of understanding of query languages such as SQL. We propose Seq2SQL, a deep neural network for translating natural language questions to corresponding SQL queries. Our model leverages the structure of SQL queries to significantly reduce the output space of generated queries. Moreover, we use rewards from in-the-loop query execution over the database to learn a policy to generate unordered parts of the query, which we show are"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By applying policy-based reinforcement learning with a query execution environment to WikiSQL, our model Seq2SQL outperforms attentional sequence to sequence models, improving execution accuracy from 35.9% to 59.4% and logical form accuracy from 23.4% to 48.3%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That rewards obtained by executing generated queries on the database provide a sufficiently dense and stable training signal for the policy, especially for the unordered components of SQL.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Seq2SQL uses deep learning plus reinforcement learning to generate SQL from natural language, reaching 59.4% execution accuracy on the new WikiSQL dataset of 80k examples.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Seq2SQL translates natural language questions into SQL queries by combining structured generation with reinforcement learning rewards from database executions, reaching 59.4 percent execution accuracy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"02f1cab606ebe168e11c252ab744f574a9322c5c490527d61641dffcc1c5421a"},"source":{"id":"1709.00103","kind":"arxiv","version":7},"verdict":{"id":"d180eb3c-e44e-413e-96a7-b3fdb5a9211e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T16:54:59.233529Z","strongest_claim":"By applying policy-based reinforcement learning with a query execution environment to WikiSQL, our model Seq2SQL outperforms attentional sequence to sequence models, improving execution accuracy from 35.9% to 59.4% and logical form accuracy from 23.4% to 48.3%.","one_line_summary":"Seq2SQL uses deep learning plus reinforcement learning to generate SQL from natural language, reaching 59.4% execution accuracy on the new WikiSQL dataset of 80k examples.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That rewards obtained by executing generated queries on the database provide a sufficiently dense and stable training signal for the policy, especially for the unordered components of SQL.","pith_extraction_headline":"Seq2SQL translates natural language questions into SQL queries by combining structured generation with reinforcement learning rewards from database executions, reaching 59.4 percent execution accuracy."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1709.00103/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":43,"sample":[{"doi":"","year":1995,"title":"I. Androutsopoulos, G.D. Ritchie, and P. Thanisch. Natural language interfaces to databases - an introduction. 1995","work_id":"b33e946d-1f29-43dc-9388-9de4e63852bd","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2011,"title":"Yoav Artzi and Luke S. Zettlemoyer. Bootstrapping semantic parsers from conversations. In EMNLP, 2011","work_id":"8cec51c2-7aed-4594-9bca-c207c75212e7","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2013,"title":"Yoav Artzi and Luke S. Zettlemoyer. Weakly supervised learning of semantic parsers for mapping instructions to actions. TACL, 1: 0 49--62, 2013","work_id":"56b3e86a-36ce-43ce-b295-2d61c2794fe4","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2000,"title":"A new database on the structure and development of the financial sector","work_id":"aa94016f-5f82-45af-9cc7-6f5d64da7964","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Le, Mohammad Norouzi, and Samy Bengio","work_id":"ee8f2221-c26b-4a37-a4cb-52171d86b086","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":43,"snapshot_sha256":"95a4ae75f9b9ca5ef3905eb4263173fde1cd546e2a69aff7d5ed0bef82e7813e","internal_anchors":1},"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":"1709.00103","created_at":"2026-07-04T22:19:56.633940+00:00"},{"alias_kind":"arxiv_version","alias_value":"1709.00103v7","created_at":"2026-07-04T22:19:56.633940+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1709.00103","created_at":"2026-07-04T22:19:56.633940+00:00"},{"alias_kind":"pith_short_12","alias_value":"I345H75Q2A5B","created_at":"2026-07-04T22:19:56.633940+00:00"},{"alias_kind":"pith_short_16","alias_value":"I345H75Q2A5BN6ZU","created_at":"2026-07-04T22:19:56.633940+00:00"},{"alias_kind":"pith_short_8","alias_value":"I345H75Q","created_at":"2026-07-04T22:19:56.633940+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":61,"internal_anchor_count":61,"sample":[{"citing_arxiv_id":"2607.06229","citing_title":"Spider 2.0-AIFunc: Extending Real-World Text-to-SQL to AI-Native SQL Workflows","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2606.23997","citing_title":"ChartWalker: Benchmarking the Cross-Chart RAG Task with Hierarchical Knowledge Graphs","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2606.23667","citing_title":"The Table Says Otherwise: Testing LLMs with Counterfactual Relational Data","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22586","citing_title":"Text2DSL: LLM-Based Code Generation for Domain-Specific Languages","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2606.12587","citing_title":"Strategic Decision Support for AI Agents","ref_index":98,"is_internal_anchor":true},{"citing_arxiv_id":"2606.09578","citing_title":"TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2606.08018","citing_title":"UniQL: Towards Dialect-Universal Benchmarking for Text-to-SQL","ref_index":185,"is_internal_anchor":true},{"citing_arxiv_id":"2606.05906","citing_title":"ACE-SQL: Adaptive Co-Optimization via Empirical Credit Assignment for Text-to-SQL","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2606.04500","citing_title":"SANE Schema-aware Natural-language Evaluation of Biological Data","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2606.03363","citing_title":"EntSQL: A Benchmark for Grounding Text-to-SQL in Long-Context Enterprise Knowledge","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02170","citing_title":"CRAFTQA: A Code-Driven Adaptive Framework for Complex Structured Data Reasoning","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2606.01007","citing_title":"Beyond Task-Agnostic: Task-Aware Grouping for Communication-Efficient Multi-Task MoE Inference","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2606.30851","citing_title":"Test-Time Verification for Text-to-SQL via Outcome Reward Models","ref_index":119,"is_internal_anchor":true},{"citing_arxiv_id":"2606.23693","citing_title":"EXPO-SQL: Execution-based Clause-level Policy Optimization for Text-to-SQL","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23595","citing_title":"Learning to Evaluate: Cost-Effective Model Evaluation on Unlabeled Data with Meta-Learning","ref_index":65,"is_internal_anchor":true},{"citing_arxiv_id":"2606.28916","citing_title":"Latent Bridges for Multi-Table Question Answering","ref_index":121,"is_internal_anchor":true},{"citing_arxiv_id":"2605.27164","citing_title":"Query Symbolically or Retrieve Semantically? A Dataset and Method for Semi-Structured Question Answering","ref_index":54,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00547","citing_title":"Learning to Retrieve: Dual-Level Long-Term Memory for Text-to-SQL Agents","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2606.03327","citing_title":"CAPER: Clause-Aligned Process Supervision for Text-to-SQL","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"1906.09302","citing_title":"Neural Machine Translating from Natural Language to SPARQL","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"1906.11790","citing_title":"Encoding Database Schemas with Relation-Aware Self-Attention for Text-to-SQL Parsers","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"1906.12348","citing_title":"MLFriend: Interactive Prediction Task Recommendation for Event-Driven Time-Series Data","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2305.07759","citing_title":"TinyStories: How Small Can Language Models Be and Still Speak Coherent English?","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2605.23595","citing_title":"Learning to Evaluate: Cost-Effective Model Evaluation on Unlabeled Data with Meta-Learning","ref_index":105,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22843","citing_title":"Knowledge Distillation for Low-Resource Open-source Text-to-SQL Model","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC","json":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC.json","graph_json":"https://pith.science/api/pith-number/I345H75Q2A5BN6ZUDPV3VWAOUC/graph.json","events_json":"https://pith.science/api/pith-number/I345H75Q2A5BN6ZUDPV3VWAOUC/events.json","paper":"https://pith.science/paper/I345H75Q"},"agent_actions":{"view_html":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC","download_json":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC.json","view_paper":"https://pith.science/paper/I345H75Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1709.00103&json=true","fetch_graph":"https://pith.science/api/pith-number/I345H75Q2A5BN6ZUDPV3VWAOUC/graph.json","fetch_events":"https://pith.science/api/pith-number/I345H75Q2A5BN6ZUDPV3VWAOUC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC/action/storage_attestation","attest_author":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC/action/author_attestation","sign_citation":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC/action/citation_signature","submit_replication":"https://pith.science/pith/I345H75Q2A5BN6ZUDPV3VWAOUC/action/replication_record"}},"created_at":"2026-07-04T22:19:56.633940+00:00","updated_at":"2026-07-04T22:19:56.633940+00:00"}