[
  {
    "path": "LICENSE",
    "content": "                                 Apache License\n                           Version 2.0, January 2004\n                        http://www.apache.org/licenses/\n\n   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION\n\n   1. Definitions.\n\n      \"License\" shall mean the terms and conditions for use, reproduction,\n      and distribution as defined by Sections 1 through 9 of this document.\n\n      \"Licensor\" shall mean the copyright owner or entity authorized by\n      the copyright owner that is granting the License.\n\n      \"Legal Entity\" shall mean the union of the acting entity and all\n      other entities that control, are controlled by, or are under common\n      control with that entity. For the purposes of this definition,\n      \"control\" means (i) the power, direct or indirect, to cause the\n      direction or management of such entity, whether by contract or\n      otherwise, or (ii) ownership of fifty percent (50%) or more of the\n      outstanding shares, or (iii) beneficial ownership of such entity.\n\n      \"You\" (or \"Your\") shall mean an individual or Legal Entity\n      exercising permissions granted by this License.\n\n      \"Source\" form shall mean the preferred form for making modifications,\n      including but not limited to software source code, documentation\n      source, and configuration files.\n\n      \"Object\" form shall mean any form resulting from mechanical\n      transformation or translation of a Source form, including but\n      not limited to compiled object code, generated documentation,\n      and conversions to other media types.\n\n      \"Work\" shall mean the work of authorship, whether in Source or\n      Object form, made available under the License, as indicated by a\n      copyright notice that is included in or attached to the work\n      (an example is provided in the Appendix below).\n\n      \"Derivative Works\" shall mean any work, whether in Source or Object\n      form, that is based on (or derived from) the Work and for which the\n      editorial revisions, annotations, elaborations, or other modifications\n      represent, as a whole, an original work of authorship. For the purposes\n      of this License, Derivative Works shall not include works that remain\n      separable from, or merely link (or bind by name) to the interfaces of,\n      the Work and Derivative Works thereof.\n\n      \"Contribution\" shall mean any work of authorship, including\n      the original version of the Work and any modifications or additions\n      to that Work or Derivative Works thereof, that is intentionally\n      submitted to Licensor for inclusion in the Work by the copyright owner\n      or by an individual or Legal Entity authorized to submit on behalf of\n      the copyright owner. For the purposes of this definition, \"submitted\"\n      means any form of electronic, verbal, or written communication sent\n      to the Licensor or its representatives, including but not limited to\n      communication on electronic mailing lists, source code control systems,\n      and issue tracking systems that are managed by, or on behalf of, the\n      Licensor for the purpose of discussing and improving the Work, but\n      excluding communication that is conspicuously marked or otherwise\n      designated in writing by the copyright owner as \"Not a Contribution.\"\n\n      \"Contributor\" shall mean Licensor and any individual or Legal Entity\n      on behalf of whom a Contribution has been received by Licensor and\n      subsequently incorporated within the Work.\n\n   2. Grant of Copyright License. Subject to the terms and conditions of\n      this License, each Contributor hereby grants to You a perpetual,\n      worldwide, non-exclusive, no-charge, royalty-free, irrevocable\n      copyright license to reproduce, prepare Derivative Works of,\n      publicly display, publicly perform, sublicense, and distribute the\n      Work and such Derivative Works in Source or Object form.\n\n   3. Grant of Patent License. Subject to the terms and conditions of\n      this License, each Contributor hereby grants to You a perpetual,\n      worldwide, non-exclusive, no-charge, royalty-free, irrevocable\n      (except as stated in this section) patent license to make, have made,\n      use, offer to sell, sell, import, and otherwise transfer the Work,\n      where such license applies only to those patent claims licensable\n      by such Contributor that are necessarily infringed by their\n      Contribution(s) alone or by combination of their Contribution(s)\n      with the Work to which such Contribution(s) was submitted. If You\n      institute patent litigation against any entity (including a\n      cross-claim or counterclaim in a lawsuit) alleging that the Work\n      or a Contribution incorporated within the Work constitutes direct\n      or contributory patent infringement, then any patent licenses\n      granted to You under this License for that Work shall terminate\n      as of the date such litigation is filed.\n\n   4. Redistribution. You may reproduce and distribute copies of the\n      Work or Derivative Works thereof in any medium, with or without\n      modifications, and in Source or Object form, provided that You\n      meet the following conditions:\n\n      (a) You must give any other recipients of the Work or\n          Derivative Works a copy of this License; and\n\n      (b) You must cause any modified files to carry prominent notices\n          stating that You changed the files; and\n\n      (c) You must retain, in the Source form of any Derivative Works\n          that You distribute, all copyright, patent, trademark, and\n          attribution notices from the Source form of the Work,\n          excluding those notices that do not pertain to any part of\n          the Derivative Works; and\n\n      (d) If the Work includes a \"NOTICE\" text file as part of its\n          distribution, then any Derivative Works that You distribute must\n          include a readable copy of the attribution notices contained\n          within such NOTICE file, excluding those notices that do not\n          pertain to any part of the Derivative Works, in at least one\n          of the following places: within a NOTICE text file distributed\n          as part of the Derivative Works; within the Source form or\n          documentation, if provided along with the Derivative Works; or,\n          within a display generated by the Derivative Works, if and\n          wherever such third-party notices normally appear. The contents\n          of the NOTICE file are for informational purposes only and\n          do not modify the License. You may add Your own attribution\n          notices within Derivative Works that You distribute, alongside\n          or as an addendum to the NOTICE text from the Work, provided\n          that such additional attribution notices cannot be construed\n          as modifying the License.\n\n      You may add Your own copyright statement to Your modifications and\n      may provide additional or different license terms and conditions\n      for use, reproduction, or distribution of Your modifications, or\n      for any such Derivative Works as a whole, provided Your use,\n      reproduction, and distribution of the Work otherwise complies with\n      the conditions stated in this License.\n\n   5. Submission of Contributions. Unless You explicitly state otherwise,\n      any Contribution intentionally submitted for inclusion in the Work\n      by You to the Licensor shall be under the terms and conditions of\n      this License, without any additional terms or conditions.\n      Notwithstanding the above, nothing herein shall supersede or modify\n      the terms of any separate license agreement you may have executed\n      with Licensor regarding such Contributions.\n\n   6. Trademarks. This License does not grant permission to use the trade\n      names, trademarks, service marks, or product names of the Licensor,\n      except as required for reasonable and customary use in describing the\n      origin of the Work and reproducing the content of the NOTICE file.\n\n   7. Disclaimer of Warranty. Unless required by applicable law or\n      agreed to in writing, Licensor provides the Work (and each\n      Contributor provides its Contributions) on an \"AS IS\" BASIS,\n      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or\n      implied, including, without limitation, any warranties or conditions\n      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A\n      PARTICULAR PURPOSE. You are solely responsible for determining the\n      appropriateness of using or redistributing the Work and assume any\n      risks associated with Your exercise of permissions under this License.\n\n   8. Limitation of Liability. In no event and under no legal theory,\n      whether in tort (including negligence), contract, or otherwise,\n      unless required by applicable law (such as deliberate and grossly\n      negligent acts) or agreed to in writing, shall any Contributor be\n      liable to You for damages, including any direct, indirect, special,\n      incidental, or consequential damages of any character arising as a\n      result of this License or out of the use or inability to use the\n      Work (including but not limited to damages for loss of goodwill,\n      work stoppage, computer failure or malfunction, or any and all\n      other commercial damages or losses), even if such Contributor\n      has been advised of the possibility of such damages.\n\n   9. Accepting Warranty or Additional Liability. While redistributing\n      the Work or Derivative Works thereof, You may choose to offer,\n      and charge a fee for, acceptance of support, warranty, indemnity,\n      or other liability obligations and/or rights consistent with this\n      License. However, in accepting such obligations, You may act only\n      on Your own behalf and on Your sole responsibility, not on behalf\n      of any other Contributor, and only if You agree to indemnify,\n      defend, and hold each Contributor harmless for any liability\n      incurred by, or claims asserted against, such Contributor by reason\n      of your accepting any such warranty or additional liability.\n\n   END OF TERMS AND CONDITIONS\n\n   APPENDIX: How to apply the Apache License to your work.\n\n      To apply the Apache License to your work, attach the following\n      boilerplate notice, with the fields enclosed by brackets \"[]\"\n      replaced with your own identifying information. (Don't include\n      the brackets!)  The text should be enclosed in the appropriate\n      comment syntax for the file format. We also recommend that a\n      file or class name and description of purpose be included on the\n      same \"printed page\" as the copyright notice for easier\n      identification within third-party archives.\n\n   Copyright [yyyy] [name of copyright owner]\n\n   Licensed under the Apache License, Version 2.0 (the \"License\");\n   you may not use this file except in compliance with the License.\n   You may obtain a copy of the License at\n\n       http://www.apache.org/licenses/LICENSE-2.0\n\n   Unless required by applicable law or agreed to in writing, software\n   distributed under the License is distributed on an \"AS IS\" BASIS,\n   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n   See the License for the specific language governing permissions and\n   limitations under the License.\n"
  },
  {
    "path": "README.md",
    "content": "## NL2LF \n___(持续更新中...)___   \n**_recently update log:_**  \n\n_0. UnifiedSKG, UniSAr_  \n_1. GNN works: LGESQL, ShadowGNN, SADGA, S²SQL (SOTA)_   \n_2. RatSQL + Pretraining (STRUG, GraPPa, GAP, GP) + NatSQL_   \n_3. PICARD, DT-Fixup, RaSaP_  \n_4. wikisql: SeaD, SeqGenSQL, BRIDGE^_   \n\n\n\n\n> The Resources for `Natural Language to Logical Form` Research, Focus on `NL2SQL` first.  \n> \"自然语言转逻辑形式\"研究资料收集: \n> 本阶段主要以**NL2SQL**的研究为主, 主要包括评测公开数据集、相关论文和部分代码实现、相关博客或公众号文章。\n\n[NL2SQL](#nl2sql--text2sql)  \n[一、主要评测数据集 dataset](#一主要评测数据集dataset)  \n[二、主要论文方法及代码实现 papers&code](#二主要论文方法及代码实现paperscode)  \n&nbsp;&nbsp;&nbsp;&nbsp;[1. WikiSQL](#1-wikisql)  \n&nbsp;&nbsp;&nbsp;&nbsp;[2. Spider](#2-spider)  \n&nbsp;&nbsp;&nbsp;&nbsp;[3. UnifiedSKG](#3-unifiedskg--)   \n[三、相关资源扩展 extend-resources](#三相关资源扩展-extend-resources)  \n&nbsp;&nbsp;&nbsp;&nbsp;[1. Related Works](#1-related-works)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.1. Pre-training](#11-pre-training--)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.2. Systems](#12-systems)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.3. Surveys](#13-surveys)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.4. Blogs](#14-blogs)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.5. Other Papers](#15-other-papers)  \n&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;[1.6. Tools](#16-tools)  \n&nbsp;&nbsp;&nbsp;&nbsp;[2. SQL2Seq](#2-sql2seq)  \n&nbsp;&nbsp;&nbsp;&nbsp;[3. 图神经网络 GNN](#3-图神经网络gnn)  \n\n\n### NL2SQL & Text2SQL\n#### 一、主要评测数据集(DataSet)\n---\n- **Academic, Advising, ATIS, Geography, Restaurants, Scholar, IMDB, Yelp, etc.**  \n  + `Blog` [http://jkk.name/text2sql-data/](http://jkk.name/text2sql-data/) \n  + `GitHub` [https://github.com/jkkummerfeld/text2sql-data](https://github.com/jkkummerfeld/text2sql-data)\n  + `Paper` [Improving Text-to-SQL Evaluation Methodology](https://www.aclweb.org/anthology/P18-1033.pdf), _Finegan-Dollak C, Kummerfeld J K, Zhang L, et al._, ACL 2018  \n---\n\n- **WikiTableQuestions**\n  + `Home`  [WikiTableQuestions: a Complex Real-World Question Understanding Dataset](https://nlp.stanford.edu/blog/wikitablequestions-a-complex-real-world-question-understanding-dataset/)\n\n---\n- **WikiSQL**  \n  > WikiSQL数据集特点:\n  > 1. 单表单列查询; \n  > 2. 聚合操作('MAX', 'MIN', 'COUNT', 'SUM', 'AVG'); \n  > 3. 条件连接('AND'); \n  > 4. 条件比较('=', '>', '<')\n  + `GitHub` [https://github.com/salesforce/WikiSQL](https://github.com/salesforce/WikiSQL)\n  + `Paper` [Seq2sql: Generating structured queries from natural language using reinforcement learning](https://arxiv.org/pdf/1709.00103.pdf), _Zhong V, Xiong C, Socher R._ , 2017.\n\n---\n- **Spider**  \n  > Spider数据集特点:\n  > 1. Complex, Cross-domain and Zero-shot \n  > 2. 多表多列查询, 复杂子查询;\n  > 3. 聚合操作('MAX', 'MIN', 'COUNT', 'SUM', 'AVG','GROUP', 'HAVING', 'LIMIT'); \n  > 4. join连接：('join', 'on', 'as')\n  > 5. where连接：('AND','OR'); \n  > 6. where操作：('not', 'between', '=', '>', '<', '>=', '<=', '!=', 'in', 'like', 'is', 'exists')\n  > 7. 排序操作：('order by', 'desc', 'asc')\n  > 8. sql连接：('Intersect', 'Union', 'Except')\n  + `Home` [https://yale-lily.github.io/spider](https://yale-lily.github.io/spider)\n  + `GitHub` [https://github.com/taoyds/spider](https://github.com/taoyds/spider)\n  + `Paper` [Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Task](https://arxiv.org/pdf/1809.08887.pdf) _Yu T, Zhang R, Yang K, et al._ , EMNLP 2018.\n  + `PPT` [spider/wikisql/tableQA数据集统计对比_by gibbsxiong](PPT/三个数据集统计分析.pptx)\n\n---\n- **SParC**\n  > SParC数据集特点:\n  > 1. Context-dependent and Multi-turn version of the Spider task.  \n       继承Spider特点的上下文多轮任务。\n  + `Home` [https://yale-lily.github.io/sparc](https://yale-lily.github.io/sparc)\n  + `Paper`[SParC: Cross-Domain Semantic Parsing in Context](https://arxiv.org/pdf/1906.02285.pdf), _Yu T, Zhang R, Yasunaga M, et al._, ACL 2019.\n\n---\n- **CoSQL**\n  > CoSQL数据集特点: \n  > 1. Cross-domain Conversational, the Dilaogue version of the Spider and SParC tasks.  \n  >    继承Spider特点的多轮对话任务，涉及意图澄清。\n  + `Home` [https://yale-lily.github.io/cosql](https://yale-lily.github.io/cosql)\n  + `Paper` [CoSQL: A Conversational Text-to-SQL Challenge Towards Cross-Domain Natural Language Interfaces to Databases](https://arxiv.org/pdf/1909.05378.pdf), _Yu T, Zhang R, Er H Y, et al._, EMNLP-IJCNLP 2019. \n\n---\n- **Chinese Spider**\n  > 中文版Spider  \n  + `Home` [https://taolusi.github.io/CSpider-explorer/](https://taolusi.github.io/CSpider-explorer/)\n  + `GitHub` [https://github.com/taolusi/chisp](https://github.com/taolusi/chisp)\n  + `Paper` [A Pilot Study for Chinese SQL Semantic Parsing](https://frcchang.github.io/pub/emnlp2019.2.pdf), _Qingkai Min, Yuefeng Shi and Yue Zhang_, EMNLP-IJCNLP 2019.\n\n---\n- **TableQA** \n  > 首届中文NL2SQL挑战赛 \n  > 数据特点:\n  > 1. 中文加强版WikiSql，金融等泛领域数据\n  > 2. 单表多列(两列)查询\n  > 3. 聚合操作('MAX', 'MIN', 'COUNT', 'SUM', 'AVG'); \n  > 4. 条件连接('AND', 'OR'); \n  > 5. 条件比较('=', '>', '<', '!=')\n  + `Home` [https://tianchi.aliyun.com/competition/entrance/231716/information](https://tianchi.aliyun.com/competition/entrance/231716/information)\n  + `GitHub` [https://github.com/ZhuiyiTechnology/nl2sql_baseline](https://github.com/ZhuiyiTechnology/nl2sql_baseline)\n  + `Paper` [TableQA: a Large-Scale Chinese Text-to-SQL Dataset for Table-Aware SQL Generation](https://arxiv.org/pdf/2006.06434.pdf)[J]. Sun N, Yang X, Liu Y. 2020.\n  + `Blog` \n    1. [人工智能时代如何高效发掘数据库的价值？NL2SQL值得你关注](https://www.jiqizhixin.com/articles/2019-06-02-3)\n    2. [让机器自动写SQL语言，首届中文NL2SQL挑战赛等你来战](https://mp.weixin.qq.com/s/wuC8O6DuxxtuFe8BO1ilZQ)\n    3. [一图读懂首届中文NL2SQL挑战赛](https://mp.weixin.qq.com/s/0FhfX0r643umZ36HxffLmQ)\n    4. [基于Bert的NL2SQL模型：一个简明的Baseline_苏剑林](https://kexue.fm/archives/6771)\n    5. [首届中文NL2SQL挑战赛圆满结束，产学研携手共推智能交互发展](https://mp.weixin.qq.com/s/F747u_n3vVC2y8Z1dOaoNw)\n  + `RANK` \n    1. [冠军方案_MSQL_Top1.pdf](PPT/Top1.pdf) \n       [https://github.com/nudtnlp/tianchi-nl2sql-top1](https://github.com/nudtnlp/tianchi-nl2sql-top1)\n    2. [亚军方案_top2.pdf](PPT/top2.pdf)\n    3. [季军方案_top3.pdf](PPT/top3.pdf)\n       第三名代码实现 [https://github.com/beader/tianchi_nl2sql](https://github.com/beader/tianchi_nl2sql)\n    4. [第四名方案_top4.pdf](PPT/top4.pdf)\n    5. [第五名方案_top5.pdf](PPT/top5.pdf)\n    6. 第六名代码实现 [https://github.com/eguilg/nl2sql](https://github.com/eguilg/nl2sql)\n  + `Paper`   \n    1. Zhang X, Yin F, Ma G, et al. [M-SQL: Multi-Task Representation Learning for Single-Table Text2sql Generation](https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9020099)[J]. IEEE Access, 2020, 8: 43156-43167. 🔥\n\n---\n- **DuSQL**  \n  > 百度2020语言与智能技术竞赛：语义解析任务，大规模开放领域的复杂中文Text-to-SQL数据集\n  > 数据特点：\n  > 1. 包含200个Database以及对应的2.3万对(question, SQL query)，其中18000对用于训练集，2000用于验证集，3000用于测试集。\n  > 2. 200个Database来自百科infobox、百科表格数据、以及互联网上存在的表格数据。每个Database包含若干张表格（2-11张，平均4.1张），人工构建了表之间的链接操作（即foreign key）。为了验证解析算法Database无关性及question无关性，在训练集合和测试集合的Database无交叉。\n  > 3. 包含复杂的多表join查询和嵌套查询，复杂度和spider类似。评价方法关注每一个组件的精准匹配度，并消除顺序影响。因此对val的准确度要求更高。具体的sql嵌套结构单元分解如下：\n\n  ``` shell\n  # 关键词和嵌套规则\n  select: [(agg_id, val_unit), (agg_id, val_unit), ...]\n  from: {'table_units': [table_unit, table_unit, ...], 'conds': condition}\n  where: condition\n  groupBy: [col_unit, ...]\n  orderBy: asc/desc, [(agg_id, val_unit), ...]\n  having: condition\n  limit: None/number\n  intersect: None/sql\n  except: None/sql \n  union: None/sql\n  ```\n  ```shell\n  # 连接单元\n  val: number(float)/string(str)/sql(dict)\n  col_unit: (agg_id, col_id)\n  val_unit: (calc_op, col_unit1, col_unit2)\n  table_type: 'table_unit'/'sql'\n  table_unit: (table_type, table_id/sql)\n  cond_unit: (agg_id, cond_op, val_unit, val1, val2)\n  condition: [cond_unit1, 'and'/'or', cond_unit2, ...]\n  ```\n  ```shell\n  # op操作符\n  agg_id: (none, max, min, count, sum, avg)\n  calc_op: (none, -, +, \\*, /)\n  cond_op: (not_in, between, =, >, <, >=, <=, !=, in, like)\n  ```\n  + `Home` [https://aistudio.baidu.com/aistudio/competition/detail/30?isFromCcf=true](https://aistudio.baidu.com/aistudio/competition/detail/30?isFromCcf=true)    \n  + `GitHub` [https://github.com/PaddlePaddle/Research/tree/master/NLP/DuSQL-Baseline](https://github.com/PaddlePaddle/Research/tree/master/NLP/DuSQL-Baseline)\n  + `Video` 冠军分享  [http://mbd.baidu.com/webpage?type=live&action=liveshow&source=h5pre&room_id=4008201814](http://mbd.baidu.com/webpage?type=live&action=liveshow&source=h5pre&room_id=4008201814)\n  + `Paper` \n     - [ ] Wang L, Zhang A, Wu K, et al. [DuSQL: A Large-Scale and Pragmatic Chinese Text-to-SQL Dataset](https://www.aclweb.org/anthology/2020.emnlp-main.562.pdf)[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020: 6923-6935.  \n  + `Dataset`\n     [https://github.com/luge-ai/luge-ai/blob/master/semantic-parsing/semantic-parsing.md](https://github.com/luge-ai/luge-ai/blob/master/semantic-parsing/semantic-parsing.md)\n  +  `Blog`  \n     1. [“语义解析”任务FAQ](https://mp.weixin.qq.com/s/EXS1dgGzAyttAeHwWjrt2A)   \n     2. [语义解析 (Text-to-SQL) 技术研究及应用 上篇](https://mp.weixin.qq.com/s/FtsA4O_VTUqhhYS3Gq3G8Q)  \n     3. [语义解析 (Text-to-SQL) 技术研究及应用 下篇](https://mp.weixin.qq.com/s/5lTLW5OOuRMo2zjbzMxr_Q)  \n     4. [UNIT结构化知识问答产品](https://ai.baidu.com/forum/topic/show/957042)  \n\n---\n- **CCKS2022：金融NL2SQL评测**\n  > 现有NL2SQL数据和方法主要关注“封闭场景指定库/表”设定，这很难满足业务范围动态发展的需求。从领域特性来看，金融数据多为时间序列，包括日频行情、季频财报、年度GDP、不定期股票质押解质押等，这无疑会增大问题转SQL难度。\n\n+ `Home` [https://www.biendata.xyz/competition/fin_nl2sql/](https://www.biendata.xyz/competition/fin_nl2sql/)\n+ `DOC` [评测任务书-金融NL2SQL评测任务](http://sigkg.cn/ccks2022/wp-content/uploads/2022/04/CCKS2022-%E9%87%91%E8%9E%8DNL2SQL.docx)\n\n\n#### 二、主要论文方法及代码实现（Papers&Code）\n> 论文主要以WikiSQL和Spider为评测数据，相应排行榜详见任务主页。  \n> 下面主要整理具有代表性的方法，持续更新补充...  \n> 注: \n> Exe_score 表示 | model | Dev accuracy |  Test accuracy  |，表示执行准确率(Execution accuracy)  \n> Log_score 表示逻辑准确率(Logical accuracy)，且Spider中不包括值预测。\n\n#### **`1. WikiSQL:`** \n- **`Weakly Supervised`**  \n  > 采用弱监督方法，即不使用sql的逻辑形式作为监督信号。 \n\n  `Paper`\n  - [ ] Li N, Keller B, Butler M, et al. [SeqGenSQL--A Robust Sequence Generation Model for Structured Query Language](https://arxiv.org/pdf/2011.03836.pdf)[J]. 2020. 🔥\n  - [ ] Min S, Chen D, Hajishirzi H, et al. [A discrete hard em approach for weakly supervised question answering](https://www.cs.princeton.edu/~danqic/papers/emnlp2019.pdf)[C]. EMNLP-IJCNLP 2019.  \n  - [ ] Wang B, Titov I, Lapata M. [Learning Semantic Parsers from Denotations with Latent Structured Alignments and Abstract Programs](https://arxiv.org/pdf/1909.04165.pdf). 2019.\n  - [ ] Agarwal R, Liang C, Schuurmans D, et al. [Learning to Generalize from Sparse and Underspecified Rewards](https://arxiv.org/pdf/1902.07198.pdf). 2019.  \n  - [ ] Liang C, Norouzi M, Berant J, et al. [Memory augmented policy optimization for program synthesis and semantic parsing](https://papers.nips.cc/paper/8204-memory-augmented-policy-optimization-for-program-synthesis-and-semantic-parsing.pdf)[C].NeurIPS, 2018: 9994-10006.\n  - [ ] Guo T, Gao H. [Using Database Rule for Weak Supervised Text-to-SQL Generation](https://arxiv.org/pdf/1907.00620.pdf)[J]. 2019.\n\n\n  `Code` \n  - Hard-EM [https://github.com/shmsw25/qa-hard-em](https://github.com/shmsw25/qa-hard-em) 🔥 \n  - LatentAlignment [https://github.com/berlino/weaksp_em19](https://github.com/berlino/weaksp_em19)\n  - MeRL / MAPO [https://github.com/google-research/google-research/tree/master/meta_reward_learning](https://github.com/google-research/google-research/tree/master/meta_reward_learning)\n  - Rule-SQL [https://github.com/guotong1988/Rule-SQL](https://github.com/guotong1988/Rule-SQL)\n\n  `Exe_score`  \n\n  |Hard-EM|84.4 |  83.9  |\n  |-|-|-|\n  |LatentAlignment| 79.4 | 79.3 |\n  |MeRL | 74.9 | 74.8 |\n  |MAPO | 72.2 | 72.1 |\n  |Rule-SQL| 61.1 | 61.0 |\n\n---\n---\n- **`ExecutionGuided`**  \n  > Execution Guided (EG) 可以在解码阶段通过执行错误对生成sql的项进行修正,从而过滤了一些不符合实际的sql语句。主要分为三类执行错误：1）句法解析错误，即生成的sql语法错误。2）执行失败。常见的run-time error, 例如SUM( ) 和比较string类型的数据；3）假设执行结果不为空，则空查询的条件错误。例如条件值实际并不存在于预测的列中, 因此会去 Beam Search 实际包含条件值的列。\n\n  `Paper`  \n  - [ ]  Wang C, Huang P S, Polozov A, et al. [Robust Text-to-SQL Generation with Execution-Guided Decoding](https://arxiv.org/pdf/1807.03100.pdf)[J]. 2018.  \n  - [ ]  Wang C, Brockschmidt M, Singh R. [Pointing out SQL queries from text](https://openreview.net/pdf?id=BkUDW_lCb)[J]. 2018. \n  - [ ]  Dong L, Lapata M. [Coarse-to-fine decoding for neural semantic parsing](https://arxiv.org/pdf/1805.04793.pdf)[J]. 2018.\n  - [ ]  Huang P S, Wang C, Singh R, et al. [Natural language to structured query generation via meta-learning](https://arxiv.org/pdf/1803.02400.pdf)[J]. 2018.\n\n\n  `Code`  \n  - [https://github.com/microsoft/PointerSQL](https://github.com/microsoft/PointerSQL)  \n  - [https://github.com/donglixp/coarse2fine](https://github.com/donglixp/coarse2fine)  \n\n  `Exe_score`  \n\n  |Coarse2Fine + EG| 84.0 | 83.8 |\n  |-|-|-|\n  | Coarse2Fine | 79.0 | 78.5 |\n  | Pointer-SQL + EG | 78.4 |  78.3  |\n  | Pointer-SQL | 72.5 | 71.9 |\n\n---\n---\n\n- **`SQLNet Framework`**  \n  > 设计了一种满足SQL语法的框架, 在这样的语法框架内，只需要预测并填充相应的槽位。\n  > 语法框架为：\n  > ```\n  > SELECT $AGG $COLUMN\n  > WHERE $COLUMN $OP $VALUE\n  > (AND $COLUMN $OP $VALUE)*\n  > ```\n  > 在这基础上去完成不同的联合任务的分类预测：\n  > 1) select-column, 选择的列\n  > 2) select-aggregation， 聚合操作类型\n  > 3) where-number， where条件语句的数量\n  > 4) where-column， where条件中的列\n  > 5) where-operator， where条件操作类型（'<','=','>'）\n  > 6) where-value， where条件值\n\n  `Paper`  \n  - [ ]  Xu X, Liu C, Song D. [SQLNet: Generating structured queries from natural language without reinforcement learning](https://arxiv.org/pdf/1711.04436.pdf)[J]. 2018.  \n  - [ ]  Hwang W, Yim J, Park S, et al. [A Comprehensive Exploration on WikiSQL with Table-Aware Word Contextualization](https://arxiv.org/pdf/1902.01069.pdf)[J]. 2019.  \n  - [ ]  He P, Mao Y, Chakrabarti K, et al. [X-SQL: reinforce schema representation with context](https://arxiv.org/pdf/1908.08113.pdf)[J]. 2019. 🔥\n  - [ ]  Tong Guo, Huilin Gao. [Content Enhanced BERT-based Text-to-SQL Generation ](https://arxiv.org/pdf/1910.07179.pdf).2019.\n  - [ ]  Qin Lyu, Kaushik Chakrabarti, Shobhit Hathi, Souvik Kundu,\n  Jianwen Zhang, Zheng Chen. [Hybrid Ranking Network for Text-to-SQL](https://www.microsoft.com/en-us/research/uploads/prod/2020/03/HydraNet_20200311-5e69612887fcb.pdf). 2020 🔥\n\n  `Code`  \n  - [https://github.com/naver/sqlova](https://github.com/naver/sqlova)  \n  - [https://github.com/xiaojunxu/SQLNet](https://github.com/xiaojunxu/SQLNet)  \n  - [https://github.com/guotong1988/NL2SQL-BERT](https://github.com/guotong1988/NL2SQL-BERT)\n\n  `Exe_score`  \n\n  | RoBERTa-Large-HydraNet + EG |92.4|92.2|\n  |-|-|-|\n  | BERT-Large-HydraNet + EG |92.2|91.8|\n  | RoBERTa-Large-HydraNet |89.1|89.2|\n  | BERT-Large-HydraNet |88.9 | 88.6 |\n  | BERT-XSQL-Attention + EG |92.3|91.8|\n  | (Tong) BERT-base-TableContent-used + EG| 91.1 | 90.1 |\n  | (Tong) BERT-base-TableContent-used| 90.3 | 89.2 |\n  | BERT-XSQL-Attention | 89.5 | 88.7 |\n  | BERT-SQLova-LSTM|87.2 |  86.2  |\n  |BERT-SQLova-LSTM + EG | 90.2 | 89.6 |\n  |GloVe-SQLNet-BiLSTM|69.8 |  68.0  |\n\n---\n---\n\n- **`Schema aware Denoising (SeaD)`**  🔥🔥\n  > 在text-to-SQL任务中，由于架构设计的限制，seq2seq模型通常会导致局部最优。在本文中，作者提出了一种简单而有效的方法：采用基于transformer的seq2seq模型来加强文本到SQL生成。使用模式感知去噪（SeaD）对seq2seq模型进行训练：由两个去噪目标组成，训练模型从erosion和随机噪声中恢复输入或预测输出(自回归方式)，而不是对encoder施加约束或将任务重新格式化为槽位填充。这些去噪目标作为辅助任务，用于在seq2seq生成中更好地建模结构数据。此外，作者改进并提出了一种子句敏感执行引导（Execution Guided, EG）解码策略，以克服生成模型EG解码的局限性。\n\n  `Paper`\n  - [ ] [1] Xuan K ,  Wang Y ,  Wang Y , et al. [SeaD: End-to-end Text-to-SQL Generation with Schema-aware Denoising\n](https://arxiv.org/pdf/2105.07911.pdf)[J].  2021.\n\n  `Exe_score`  \n\n  | SeaD + EG |92.9|93.0|\n  |-|-|-|\n  | SeaD |90.2|90.1|\n\n---\n---\n\n- **`Schema Dependency Guided`**  🔥🔥\n  > 结合Question和Schema之间的依存关系来进行多任务学习。\n\n  `Paper`\n  - [ ] Hui B, Shi X, Geng R, et al. [Improving Text-to-SQL with Schema Dependency Learning](https://arxiv.org/pdf/2103.04399.pdf)[J]. arXiv preprint arXiv:2103.04399, 2021.\n\n  `Exe_score`  \n\n  | SDSQL + EG |92.5|92.4|\n  |-|-|-|\n  | SDSQL |88.7|88.8|\n\n---\n---\n\n- **`BRIDGE^`**   🔥\n\n  `Paper`\n  - [ ] Lin X V, Socher R, Xiong C. [Bridging Textual and Tabular Data for Cross-Domain Text-to-SQL Semantic Parsing](https://www.aclweb.org/anthology/2020.findings-emnlp.438.pdf)[C]//EMNLP: Findings. 2020: 4870-4888.\n\n  `Code` \n  - [https://github.com/salesforce/TabularSemanticParsing](https://github.com/salesforce/TabularSemanticParsing)\n  - [https://github.com/WING-NUS/slsql](https://github.com/WING-NUS/slsql)\n\n  `Exe_score`  \n\n  | BRIDGE^ + EG |92.6|91.9|\n  |-|-|-|\n  | BRIDGE |91.7|91.1|\n\n---\n---\n\n- **`T5 SeqGenSQL`**  🔥🔥\n  > 利用T5预训练语言（文本生成）模型, 将问题直接转换为SQL语句。同时，探索了如何利用表格模式信息对问题进行扩充，生成新的(silver)训练数据集\n\n  `Paper`\n  - [ ] Li N ,  Keller B ,  Butler M , et al. [SeqGenSQL -- A Robust Sequence Generation Model for Structured Query Language](https://arxiv.org/pdf/2011.03836.pdf)[J].  2020.\n  - [ ] Youssef M, Abdelkader R, et al.[\nSQL Generation from Natural Language: A Sequence-to-Sequence Model Powered by the Transformers Architecture and Association Rules](https://pdfs.semanticscholar.org/b877/233410484b2ff2add278105c53b6633d9d20.pdf)[J]. 2021\n\n  `Exe_score`  \n\n  |SeqGenSQL + EG |90.8|90.5|\n  |-|-|-|\n  |SeqGenSQL(T5-base + 250K silver data) |90.6|90.3|\n  |T5-large&mT5-large + Association Rules *| 91.2 | 91.0 |\n\n---\n---\n\n- **`Information Extraction Approach`**  \n  > 信息抽取的方法: 采用统一的基于BERT的抽取模型来识别query提及的槽位类型，包括序列标注方法、关系抽取和基于文本匹配的链接方法。\n\n  `Paper`\n  - [ ] Ping An Life, AI Team. [IE-SQL: Mention Extraction and Linking for SQL Query Generation](https://drive.google.com/file/d/1t3xEltqKpYJGYekAhQ5vYFen1ocHJ3sY/view) 2020\n\n  `Exe_score`  \n\n  | BERT-IE-SQL + EG |92.6|92.5|\n  |-|-|-|\n  | BERT-IE-SQL |88.7|88.8|\n\n---\n---\n\n- **`MRC Approach`**  🔥\n  > 阅读理解的方法: 与传统槽位填充方法不同的是，该方法将NL2SQL转化为QA问题,通过统一的MRC框架来预测不同的槽位。\n\n  `Paper`\n  - [ ] Yan Z, Ma J, Zhang Y, et al. [SQL Generation via Machine Reading Comprehension](https://www.aclweb.org/anthology/2020.coling-main.31.pdf)[C]//Proceedings of the 28th International Conference on Computational Linguistics. 2020: 350-356.\n  \n  `Code`  \n  - [https://github.com/nl2sql/QA-SQL](https://github.com/nl2sql/QA-SQL)  \n  \n  `Exe_score`\n\n  | BERT-MRC-SQL + STILTs training + AGG enhancement |87.8|87.4|\n  |-|-|-|\n  | BERT-MRC-SQL + STILTs training |86.2|86.0|\n  | BERT-MRC-SQL |85.9|85.9|\n\n---\n---\n\n- **`Model Interactive`**  \n  > 基于用户交互的语义解析，更偏向于落地实践。在生成sql后，通过自然语句生成来进一步要求用户进行意图澄清，从而对sql进行修正。\n\n  `Blog` \n  - [Facebook提出全新交互式语义分析框架，自然语言生成SQL语句准确率提升10%](https://mp.weixin.qq.com/s/B3Rw-Rqy8b4-HtWPBtEI_w)  \n\n  `Paper`\n  - [ ] Yao Z, Su Y, Sun H, et al. [Model-based Interactive Semantic Parsing: A Unified Framework and A Text-to-SQL Case Study](https://arxiv.org/pdf/1910.05389.pdf)[C]. EMNLP-IJCNLP 2019.  \n\n  `Code`\n  - [https://github.com/sunlab-osu/MISP](https://github.com/sunlab-osu/MISP)\n\n----\n----\n#### **`2. Spider:`**  \n\n- **`GNN Encoding Seq2Seq`**  🔥\n  > `Schema-GNN:` 利用多表关联信息来建立一个表名、列名为节点，表内、表间关系为边的图。通过GNN方法计算每一个节点(table item)的隐藏状态。在seq2seq模型的encoding阶段，每个query word 向量对每个 table item隐藏向量进行attention计算， 并将attention权重作为每个query word的图表示。在decoding阶段，结合语法规则，如果输出应为table item,则将输出向量与所有table item隐藏向量进行全连接打分，计算其关联程度。\n\n  > `LGESQL: ` 以往的建图方式存在问题：1）忽略了边在拓扑结构中丰富的语义信息 2）无法区分每个节点的局部和非局部的关系。本文方法(Line Graph Enhanced Text-toSQL)在不构建元路径的情况挖掘潜在的关系特征.借助Line Graph，消息可以有效的在连接节点之间以及拓扑有向边上进行传播。在图迭代过程中，局部和非局部关系被显著地集成。同时，还设计了图剪枝的辅助任务，来提高编码器的识别能力。\n\n  > `ShadowGNN: ` 在跨域结构下，传统的语义解析模型难以适应不可见的数据库模式。为了提高稀少且不可见模式的模型泛化能力，我们提出了一种新的架构ShadowGNN，它可以在抽象和语义级别处理schemas。具体地，通过忽略数据库中语义项的名称，抽象schemas利用图映射神经网络来获得question和schema的去符号化表示。在领域无关表示的基础上，利用关系感知转换器进一步提取question和schema之间的逻辑联系。最后，还应用了一个具有上下文无关语法的SQL解码器。\n\n  > `SADGA: ` Structure-Aware Dual Graph Aggregation Network, 设计了一种基于图结构的聚合方法来学习question图和schema图的映射关系。该聚合方法的特征来源于图的全局链接、局部链接以及双图聚合机制。\n\n  > `S²SQL: ` 以往的基于图的编码器，没有很好的建模question的句法结构。本文利用句法解析器来抽取question的信息，并将句法信息注入到question-schema图编码器中。同时还使用了解耦约束来引导不同的边关系嵌入，从而提升网络性能。\n\n  `Paper`\n  - [ ] Krishnamurthy J, Dasigi P, Gardner M. [Neural semantic parsing with type constraints for semi-structured tables](https://www.aclweb.org/anthology/D17-1160.pdf)[C]. EMNLP 2017.\n  - [ ] Lin K, Bogin B, Neumann M, et al. [Grammar-based Neural Text-to-SQL Generation](https://arxiv.org/pdf/1905.13326.pdf). 2019.\n  - [ ] Bogin B, Gardner M, Berant J. [Representing Schema Structure with Graph Neural Networks for Text-to-SQL Parsing](https://arxiv.org/pdf/1905.06241.pdf)[C]. ACL 2019.  \n  - [ ] Bogin B, Gardner M, Berant J. [Global Reasoning over Database Structures for Text-to-SQL Parsing](https://arxiv.org/pdf/1908.11214.pdf)[C]. EMNLP-IJCNLP 2019.\n  - [ ] Shaw P, Massey P, Chen A, et al. [Generating Logical Forms from Graph Representations of Text and Entities](https://arxiv.org/pdf/1905.08407.pdf)[C]. ACL 2019.\n  - [ ] Kelkar A, Relan R, Bhardwaj V, et al. [Bertrand-DR: Improving Text-to-SQL using a Discriminative Re-ranker](https://arxiv.org/pdf/2002.00557.pdf)[J]. arXiv preprint arXiv:2002.00557, 2020. \n  - [ ] Cao R ,  Chen L ,  Chen Z , et al. [LGESQL: Line Graph Enhanced Text-to-SQL Model with Mixed Local and Non-Local Relations](https://arxiv.org/abs/2106.01093)[C]. ACL. 2021.\n  - [ ] Chen Z ,  Chen L ,  Zhao Y , et al. [ShadowGNN: Graph Projection Neural Network for Text-to-SQL Parser](https://arxiv.org/pdf/2104.04689.pdf)[C]. NAACL. 2021.\n  - [ ] Cai R ,  Yuan J ,  Xu B , et al. SADGA: Structure-Aware Dual Graph Aggregation Network for Text-to-SQL(https://arxiv.org/abs/2111.00653)[C].  NeurIPS 2021.\n  - [ ] Hui B ,  Geng R ,  Wang L , et al. [S²SQL: Injecting Syntax to Question-Schema Interaction Graph Encoder for Text-to-SQL Parsers](https://arxiv.org/abs/2203.06958)[C].  ACL Findings. 2022.\n\n  `Code`  \n  - [https://github.com/benbogin/spider-schema-gnn](https://github.com/benbogin/spider-schema-gnn)\n  - [https://github.com/benbogin/spider-schema-gnn-global](https://github.com/benbogin/spider-schema-gnn-global)\n  - [https://github.com/amolk/Bertrand-DR](https://github.com/amolk/Bertrand-DR)\n  - [https://github.com/rhythmcao/text2sql-lgesql](https://github.com/rhythmcao/text2sql-lgesql) \n  - [https://github.com/WowCZ/shadowgnn](https://github.com/WowCZ/shadowgnn)\n  - [https://github.com/DMIRLAB-Group/SADGA](https://github.com/DMIRLAB-Group/SADGA)\n  - \n\n  `Log_score`\n\n  | S²SQL + ELECTRA (DB content used)|76.4|72.1|\n  |:-:|:-:|:-:|\n  | SADGA + GAP (DB content used) |73.1 | 70.1|\n  | LGESQL + ELECTRA (DB content used) | 75.1| 72.0 |\n  | LGESQL + BERT (DB content used | 74.1 | 68.3 |\n  | LGESQL + Glove (DB content used)  |67.6|62.8|\n  | ShadowGNN + RoBERTa (DB content used) |72.3|66.1|\n  | ShadowGNN (DB content used)|-|64.8|\n  | GNN + Bertrand-DR | 57.9 | 54.6 |\n  | Global-GNN  |52.7|47.4|\n  | GNN | 40.7 | 39.4 |\n  | GNN w/edge vectors| 32.1 | - |\n\n---\n---\n\n- **`RATSQL related works`**  🔥🔥🔥  \n\n  `Paper`  \n  - [ ] [1] Wang B, Shin R, Liu X, et al.[RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers ](https://arxiv.org/pdf/1911.04942.pdf)[C]. ACL 2020. \n    <!-- [ICLR 2020](https://openreview.net/forum?id=H1egcgHtvB). -->\n  - [ ] [2] Deng X, Awadallah A H, Meek C, et al. [Structure-Grounded Pretraining for Text-to-SQL](https://arxiv.org/pdf/2010.12773.pdf)[C]. NAACL, 2021.\n  - [ ] [3] Gan Y ,  Chen X ,  Xie J , et al. [Natural SQL: Making SQL Easier to Infer from Natural Language Specifications](https://arxiv.org/abs/2109.05153)[C]. EMNLP Findings. 2021.\n  - [ ] [4] Yu T, Wu C S, Lin X V, et al. [GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing](https://arxiv.org/abs/2009.13845)[C]. ICLR 2021.\n  - [ ] [5] Shi P ,  Ng P ,  Wang Z , et al. [Learning Contextual Representations for Semantic Parsing with Generation-Augmented Pre-Training](https://arxiv.org/abs/2012.10309)[C]. AAAI 2021.\n  - [ ] [6] Zhao L, Cao H, Zhao Y. [GP: Context-free Grammar Pre-training for Text-to-SQL Parsers](https://arxiv.org/pdf/2101.09901.pdf)[J]. arXiv preprint arXiv:2101.09901, 2021.\n\n    \n  `Code`\n  - [https://github.com/Microsoft/rat-sql](https://github.com/Microsoft/rat-sql)\n  - [https://github.com/ygan/NatSQL](https://github.com/ygan/NatSQL)\n  - [https://github.com/awslabs/gap-text2sql](https://github.com/awslabs/gap-text2sql)\n  - \n\n  `Exe_score`\n\n  | [3] RATSQL + GAP + NatSQL (DB content used) | 73.3 |\n  |:-:|:-:|\n  \n  `Log_score` \n\n  | RAT-SQL + GraPPa + Adv (DB content used)|75.5|70.5|\n  |:-:|:-:|:-:|\n  | RATSQL++ + ELECTRA (DB content used)|75.7|70.3|\n  | [6] RATSQL + GraPPa + GP (DB content used)|72.8|69.8|\n  | [5] RATSQL + GAP (DB content used) |71.8|69.7|\n  | [4] RATSQL + GraPPa (DB content used) |73.4|69.6|\n  | [3] RATSQL + GAP + NatSQL (DB content used) | - | 68.7 |\n  | [2] RAT-SQL + STRUG (DB content used) | 72.6 |68.4|\n  | [1] RATSQL v3 + BERT (DB content used) |69.7|65.6|\n  | [1] RATSQL v2 + BERT (DB content used) | 65.8 | 61.9 |\n  | [1] RATSQL v2 (DB content used)| 62.7| 57.2|\n  | [1] RATSQL + BERT | 60.8 | 55.7 |\n  | [1] RATSQL  |60.6|53.7|\n\n----\n----\n\n- **`MSRA: IRNet related works`**  🔥🔥  \n\n  `Blog & Video`  \n  - [智能数据分析技术，解锁Excel“对话”新功能 Conversational Data Analysis](https://www.msra.cn/zh-cn/news/features/conversational-data-analysis)\n  - [Use Ideas in Excel to get Immediate answers with ONE Click]( https://www.youtube.com/watch?v=bey_1SUTB4k)\n\n  `Paper`  \n  - [ ] Guo J, Zhan Z, Gao Y, et al. [Towards Complex Text-to-SQL in Cross-Domain Database with Intermediate Representation](https://arxiv.org/pdf/1905.08205.pdf)[C]. ACL 2019.\n  - [ ] Dong Z, Sun S, Liu H, et al. [Data-Anonymous Encoding for Text-to-SQL Generation](https://www.aclweb.org/anthology/D19-1543.pdf)[C] EMNLP-IJCNLP 2019.\n  - [ ] Liu H, Fang L, Liu Q, et al. [Leveraging Adjective-Noun Phrasing Knowledge for Comparison Relation Prediction in Text-to-SQL](https://www.aclweb.org/anthology/D19-1356.pdf)[C]. EMNLP-IJCNLP 2019.\n  - [ ] Liu Q, Chen B, Lou J G, et al. [FANDA: A Novel Approach to Perform Follow-up Query Analysis](https://arxiv.org/pdf/1901.08259.pdf)[C]. AAAI 2019.\n  - [ ] Liu Q, Chen B, Liu H, et al. [A Split-and-Recombine Approach for Follow-up Query Analysis](https://www.aclweb.org/anthology/D19-1535.pdf)[C]. EMNLP-IJCNLP 2019.\n\n    \n  `Code`\n  - [https://github.com/microsoft/IRNet](https://github.com/microsoft/IRNet)\n  - [https://github.com/neeraj-bhat/IRNet/tree/dev](https://github.com/neeraj-bhat/IRNet/tree/dev)\n\n  \n  `Log_score` \n\n  | IRNet++ + XLNet (DB content used) |65.5|60.1|\n  |:-:|:-:|:-:|\n  | IRNet++ + XLNet (DB content used) |65.5|60.1|\n  | IRNet-v2 + BERT  |63.9|55.0|\n  | IRNet + BERT-Base | 61.9 | 54.7 |\n  | IRNet-v2 | 55.4 | 48.5 |\n  | IRNet| 53.2 | 46.7 | \n\n-------\n-------\n- **`MSRA DKI Group's works`**  🔥🔥  \n\n  `Paper & Code`\n  - [https://github.com/microsoft/ContextualSP](https://github.com/microsoft/ContextualSP)\n\n  `Log_score` \n\n  | ETA + BERT (DB content used) |70.8|65.3|\n  |:-:|:-:|:-:|\n\n-----\n-----\n- **`PICARD`**\n\n  `Paper`\n  - [ ] Scholak T ,  Schucher N ,  Bahdanau D . [PICARD: Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models](https://arxiv.org/abs/2109.05093)[C].  EMNLP. 2021.\n\n  `Code`\n  - [https://github.com/ElementAI/picard](https://github.com/ElementAI/picard)\n\n  `Log_score`  \n\n  | PICARD + T5-3B (DB content used)|75.5|71.9|\n  |:-:|:-:|:-:|\n\n  `Exe_score` \n  | PICARD + T5-3B (DB content used)| - | 75.1 |\n  |:-:|:-:|:-:|\n\n-----\n-----\n- **`DT-Fixup SQL-SP`**\n\n  `Paper`\n  - [ ] Xu P ,  Kumar D ,  Yang W , et al. [Optimizing Deeper Transformers on Small Datasets](https://arxiv.org/abs/2012.15355)[C]. ACL. 2021.\n\n  `Code`\n  - [https://github.com/BorealisAI/DT-Fixup](https://github.com/BorealisAI/DT-Fixup)\n\n  `Log_score`  \n\n  | DT-Fixup SQL-SP + RoBERTa (DB content used)|75.0|70.9|\n  |:-:|:-:|:-:|\n\n-----\n-----\n- **`RaSaP`**\n\n  `Paper`\n  - [ ] Hua Ng  J ,  Wa Ng  Y ,  Wa Ng  Y , et al. [Relation Aware Semi-autoregressive Semantic Parsing for NL2SQL](https://arxiv.org/abs/2108.00804)[J].  2021.\n\n  `Log_score`  \n\n  | RaSaP + ELECTRA (DB content used) |74.7|69.0|\n  |:-:|:-:|:-:|\n\n\n------\n------\n\n- **`EditSQL`**\n\n  `Paper`\n  - [ ] Zhang R, Yu T, Er H Y, et al. [Editing-Based SQL Query Generation for Cross-Domain Context-Dependent Questions](https://arxiv.org/pdf/1909.00786.pdf)[C]. EMNLP-IJCNLP 2019.\n\n  `Code`\n  - [https://github.com/ryanzhumich/editsql](https://github.com/ryanzhumich/editsql)\n\n  `Log_score`  \n\n  | EditSQL + BERT  |57.6|53.4|\n  |:-:|:-:|:-:|\n  | EditSQL | 36.4 | 32.9 |  \n\n----\n----\n\n- **`RYANSQL`**\n\n  `Paper`\n  - [ ] Choi D H, Shin M C, Kim E G, et al. [RYANSQL: Recursively Applying Sketch-based Slot Fillings for Complex Text-to-SQL in Cross-Domain Databases]([RYANSQL: Recursively Applying Sketch-based Slot Fillings for Complex Text-to-SQL in Cross-Domain Databases](https://arxiv.org/pdf/2004.03125.pdf))[J]. 2020.\n\n  `Log_score`  \n  | RYANSQL v2 + BERT  | 70.6 | 60.6 |\n  |:-:|:-:|:-:|\n  | RYANSQL + BERT | 66.6 | 58.2 | \n  | RYANSQL  | 43.3 | - |\n\n----\n----\n\n- **`SmBoP`**   \n  > 与自上而下的自回归分析相比，半自回归自底向上解析器具有多种优势。首先，由于每个解码步骤中的子树都是并行生成的，因此理论上的运行时间是对数而不是线性复杂度。其次，自下而上的方法学习在每个步骤上学习语义子程序的表示，而不是语义上模糊的部分树。最后，SMBOP基于Transformer的层将子树相互关联起来，与传统的beam-search不同，以探索过的其他树木为条件为树进行评分。\n\n  `Paper`\n  - [ ] Rubin O, Berant J. [SmBoP: Semi-autoregressive Bottom-up Semantic Parsing](https://arxiv.org/pdf/2010.12412.pdf)[C]. NAACL, 2021.  \n\n  `Code`  \n  [https://github.com/OhadRubin/SmBop](https://github.com/OhadRubin/SmBop)\n\n  `Log_score`  \n\n  | SmBoP + GraPPa (DB content used) |74.7 | 69.5|\n  |:-:|:-:|:-:|\n  | SmBoP + BART | 66.0 | 60.5 |\n\n  `Exe_score` \n  | SmBoP + GraPPa (DB content used) | - | 71.1 |\n  |:-:|:-:|:-:|\n\n----\n----\n\n- **`SLSQL`**   \n  > Schema Linking is the crux for the current text-to-SQL task. \n\n  `Paper`\n  - [ ] Lei W, Wang W, Ma Z, et al. [Re-examining the Role of Schema Linking in Text-to-SQL](https://www.aclweb.org/anthology/2020.emnlp-main.564.pdf)[C]. EMNLP 2020: 6943-6954.\n\n  `Code` \n  - [https://github.com/salesforce/TabularSemanticParsing](https://github.com/salesforce/TabularSemanticParsing)\n  - [https://github.com/WING-NUS/slsql](https://github.com/WING-NUS/slsql)\n\n  `Log_score`  \n\n  | SLSQL + BERT + Data Annotation | 60.8 | 55.7 |\n  |:-:|:-:|:-:|\n\n----\n----\n\n- **`BRIDGE `** \n  > \n\n  `Paper`\n  - [ ] Lin X V, Socher R, Xiong C. [Bridging Textual and Tabular Data for Cross-Domain Text-to-SQL Semantic Parsing](https://www.aclweb.org/anthology/2020.findings-emnlp.438.pdf)[C]//EMNLP: Findings. 2020: 4870-4888.\n\n  `Code` \n  - [https://github.com/salesforce/TabularSemanticParsing](https://github.com/salesforce/TabularSemanticParsing)\n  - [https://github.com/WING-NUS/slsql](https://github.com/WING-NUS/slsql)\n\n\n  `Log_score`  \n\n  | BRIDGE(k = 2) + BERT (DB content used) | 65.5 | 59.2 |\n  |:-:|:-:|:-:|\n  | BRIDGE(k = 1) + BERT (DB content used) | 65.3 | - |\n\n  `Exe_score` \n  | BRIDGE v2 + BERT(ensemble) (DB content used) | - | 68.3 |\n  |:-:|:-:|:-:|\n  | BRIDGE v2 + BERT (DB content used) | - | 64.3 |\n  | BRIDGE(k = 2) + BERT (DB content used) | - | 59.9 |\n----\n----\n\n- **`GAZP `** \n  > GAZP combines a forward semantic parser with a backward utterance generator to synthesize data (e.g. utterances and SQL queries)\n  in the new environment, then selects cycleconsistent examples to adapt the parser. Unlike data-augmentation, which typically synthesizes unverified examples in the training environment, GAZP synthesizes examples in the new environment whose inputoutput consistency are verified.\n\n  `Paper`\n  - [ ] Zhong V, Lewis M, Wang S I, et al. [Grounded adaptation for zero-shot executable semantic parsing](https://arxiv.org/pdf/2009.07396.pdf)[C]. EMNLP-2020.\n\n\n  `Exe_score` \n  | GAZP + BERT | - | 53.5 |\n  |:-:|:-:|:-:|\n\n----\n----\n\n- **`SQLNet Framework`**  \n\n  `Paper`\n  - [ ] [Leveraging Adjective-Noun Phrasing Knowledge for Comparison Relation Prediction in Text-to-SQL](https://www.aclweb.org/anthology/D19-1356.pdf)[C]. EMNLP 2019\n  - [ ] Yu T, Yasunaga M, Yang K, et al. [Syntaxsqlnet: Syntax tree networks for complex and cross-domaintext-to-sql task](https://arxiv.org/pdf/1810.05237.pdf)[C]. EMNLP 2018.\n  - [ ] Dongjun Lee. [Clause-Wise and Recursive Decoding for Complex and Cross-Domain Text-to-SQL Generation](https://arxiv.org/pdf/1904.08835.pdf)[C]. EMNLP 2019.\n  - [ ] Lin K, Bogin B, Neumann M, et al. [Grammar-based Neural Text-to-SQL Generation](https://arxiv.org/pdf/1905.13326.pdf). 2019.\n\n  `Code`  \n  - [https://github.com/taoyds/syntaxSQL](https://github.com/taoyds/syntaxSQL)\n\n  `Score`  \n\n  | GrammarSQL | 34.8 | 33.8 |\n  |:-:|:-:|:-:|\n  | SyntaxSQLNet + augment  |24.8|27.2|\n  | RCSQL | 28.5 | 24.3 | \n  | SyntaxSQLNet | 18.9 | 19.7 |\n  | SQLNet | 10.9 | 12.4 |\n\n---\n---\n\n#### **`3. UnifiedSKG:`**  🔥🔥\n\n\n`Blog`  \n - [ ] [结构化知识的统一建模和多任务学习](https://mp.weixin.qq.com/s/ddhZoOHa2eWKZoHESIiK3Q)\n\n`Code`  \n - [ ] [https://github.com/hkunlp/unifiedskg](https://github.com/hkunlp/unifiedskg)\n\n`Paper`\n\n - [ ] Xie T ,  Wu C H ,  Shi P , et al. [UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models](https://arxiv.org/abs/2201.05966)[J]. arXiv e-prints, 2022.\n\n\n#### 三、相关资源扩展 (extend resources)\n##### 1. Related Works  \n##### 1.1 `Pre-training`  🔥🔥🔥 \n  > jointly learns representations of natural language utterances and table schemas by leveraging generation models to generate pre-train data.\n  - [ ] Shi P ,  Ng P ,  Wang Z , et al. [GAP: Learning Contextual Representations for Semantic Parsing with Generation-Augmented Pre-Training](https://arxiv.org/abs/2012.10309)[C]. AAAI 2021.\n  \n  >  A novel weakly supervised Structure-Grounded pretraining framework (STRUG) for text-to-SQL that can effectively learn to capture text-table alignment based on a parallel text-table corpus. \n  - [ ]Deng X, Awadallah A H, Meek C, et al. [Structure-Grounded Pretraining for Text-to-SQL](https://arxiv.org/pdf/2010.12773.pdf)[C]. NAACL, 2021.\n\n  >  A new method for Text-to-SQL parsing, Grammar Pre-training (GP),is proposed to decode deep relations between question and database.\n  - [ ]Zhao L, Cao H, Zhao Y. [GP: Context-free Grammar Pre-training for Text-to-SQL Parsers](https://arxiv.org/pdf/2101.09901.pdf)[J]. arXiv preprint arXiv:2101.09901, 2021.\n\n  >  An effective pre-training approach for table semantic parsing that learns a compositional inductive bias in the joint representations of textual and tabular data.\n  - [ ] Yu T, Wu C S, Lin X V, et al. [GraPPa: Grammar-Augmented Pre-Training for Table Semantic Parsing](https://arxiv.org/abs/2009.13845)[C]. ICLR 2021.\n  \n  > this paper designs two novel pre-training objectives to impose the desired inductive bias into the learned representations for table pre-training. \n  - [ ] Qin B ,  Wang L ,  Hui B , et al. [SDCUP: Schema Dependency-Enhanced Curriculum Pre-Training for Table Semantic Parsing](https://arxiv.org/abs/2111.09486)[J]. 2021.\n\n  > table pre-training can be realized by learning a neural SQL executor over a synthetic corpus, which is obtained by automatically synthesizing executable SQL queries. \n  - [ ] Liu Q ,  Chen B ,  Guo J , et al. [TAPEX: Table Pre-training via Learning a Neural SQL Executor(https://arxiv.org/abs/2107.07653)[J].  2021.\n  \n  > A pretrained language model that jointly learns representations for NL sentences and (semi-)structured tables.\n  - [ ] Pengcheng Yin, Graham Neubig, et al. [TaBERT: Pretraining for Joint Understanding of Textual and Tabular Data](https://arxiv.org/abs/2005.08314)[C]. ACL 2020. \n  \n  > this paper designs two novel pre-training objectives to impose the desired inductive bias into the learned representations for table pre-training. \n  - [ ] Bowen Q, LiHan W, et al.[Linking-Enhanced Pre-Training for Table Semantic Parsing](https://arxiv.org/abs/2111.09486). 2021\n\n  > Adapting a semantic parser trained on a single language.\n  - [ ] Tom Sherborne, Yumo Xu, Mirella Lapata. [Bootstrapping a Crosslingual Semantic Parser](https://arxiv.org/abs/2004.02585).2020.\n\n##### 1.2 `Systems`\n  - [ ] Zeng J, Lin X V, Xiong C, et al. [Photon: A Robust Cross-Domain Text-to-SQL System](https://arxiv.org/pdf/2007.15280.pdf)[J]. 2020.\n  - [ ] Brunner U, Stockinger K. [ValueNet: A Neural Text-to-SQL Architecture Incorporating Values](https://arxiv.org/pdf/2006.00888.pdf)[J]. 2020.\n  - [ ] Elgohary A, Hosseini S, Awadallah A H. [Speak to your Parser: Interactive Text-to-SQL with Natural Language Feedback](https://arxiv.org/pdf/2006.00888)[J]. 2020.\n\n##### 1.3 `Surveys`\n  - [ ] Jovan, Martina, Frosina. [Recent Advances in SQL Query Generation: A Survey](https://arxiv.org/pdf/2005.07667.pdf)//Part of the 17th International Conference on Informatics and Information Technologies. Received best paper award. 2020.\n\n##### 1.4 `Blogs`\n  - [ ] [NL2SQL概述：一文了解NL2SQL](https://www.csuldw.com/2019/10/20/2019-10-20-nl2sql-introduction/)  \n  - [ ] [哈工大SCIR: 一文了解Text-to-SQL](https://mp.weixin.qq.com/s/ucFtWopoErtUCYDTLv2kFg)\n  - [ ] [表格问答1：简介-朴素人工智能](https://mp.weixin.qq.com/s/iIoy9cIitlaH1p_msrbYoQ) \n  - [ ] [表格问答2：模型](https://mp.weixin.qq.com/s/aXXhR4u-AvJY-h8595RJew) \n  - [ ] [表格问答完结篇：落地应用](https://mp.weixin.qq.com/s/0uEXOTfbEq86oIVnMrogRA)\n  - [ ] [ACL2020 表格预训练工作速览](https://mp.weixin.qq.com/s/AF-Wt5lKW2-oWBgdcCyWXA)\n\n##### 1.5 `Other Papers`\n  - [ ] Dhamdhere K, McCurley K S, Nahmias R, et al. [Analyza: Exploring data with conversation](https://dl.acm.org/citation.cfm?id=3025227)[C]//Proceedings of the 22nd International Conference on Intelligent User Interfaces. ACM, 2017.\n  - [ ] Chen S, San A, Liu X, et al. [A Tale of Two Linkings: Dynamically Gating between Schema Linking and Structural Linking for Text-to-SQL Parsing](https://arxiv.org/abs/2009.14809)[C]. COLING 2020.\n  -  Dou L ,  Gao Y ,  Pan M , et al. [UniSAr: A Unified Structure-Aware Autoregressive Language Model for Text-to-SQL](https://arxiv.org/abs/2203.07781)[J].  2022.\n\n##### 1.6 `Tools`\n  -  Test suite for text2sql code: [https://github.com/taoyds/test-suite-sql-eval](https://github.com/taoyds/test-suite-sql-eval)   \n  -  Test suite for text2sql paper: Zhong R, Yu T, Klein D. [Semantic Evaluation for Text-to-SQL with Distilled Test Suites](https://arxiv.org/pdf/2010.02840.pdf)[C]. EMNLP2020.\n  -  SQL Parser [https://github.com/mozilla/moz-sql-parser](https://github.com/mozilla/moz-sql-parser)\n\n##### 2. SQL2Seq  \n\n`Paper`  \n- [ ] Xu K, Wu L, Wang Z, et al. [Graph2seq: Graph to sequence learning with attention-based neural networks](https://arxiv.org/pdf/1804.00823.pdf).2018. \n- [ ] Xu K, Wu L, Wang Z, et al. [SQL-to-text generation with graph-to-sequence model](https://arxiv.org/pdf/1809.05255.pdf)[C]. EMNLP 2018.\n\n`Code`  \n- [https://github.com/IBM/SQL-to-Text](https://github.com/IBM/SQL-to-Text)\n- [https://github.com/IBM/Graph2Seq](https://github.com/IBM/Graph2Seq)\n- [https://github.com/RandolphVI/Graph2Seq](https://github.com/RandolphVI/Graph2Seq)   \n\n##### 3. 图神经网络（GNN)  \n\n`Paper`  \n- [https://github.com/thunlp/GNNPapers](https://github.com/thunlp/GNNPapers)\n- [https://github.com/IndexFziQ/GNN4NLP-Papers](https://github.com/IndexFziQ/GNN4NLP-Papers)\n- [https://github.com/nnzhan/Awesome-Graph-Neural-Networks](https://github.com/nnzhan/Awesome-Graph-Neural-Networks)\n- [https://github.com/naganandy/graph-based-deep-learning-literature](https://github.com/naganandy/graph-based-deep-learning-literature)\n- [https://github.com/svjan5/GNNs-for-NLP](https://github.com/svjan5/GNNs-for-NLP)\n\n`Code`  \n- [https://github.com/rusty1s/pytorch_geometric](https://github.com/rusty1s/pytorch_geometric)\n- [https://github.com/dmlc/dgl](https://github.com/dmlc/dgl)\n- [https://github.com/tkipf/pygcn](https://github.com/tkipf/pygcn)\n- [https://github.com/chingyaoc/ggnn.pytorch](https://github.com/chingyaoc/ggnn.pytorch)\n- [https://github.com/bdqnghi/ggnn.tensorflow](https://github.com/bdqnghi/ggnn.tensorflow)\n\n\n"
  }
]