In sophisticated existing Text-to-SQL methods exhibit errors in various
proportions, including schema-linking errors (incorrect columns, tables, or
extra columns), join errors, nested errors, and group-by errors. Consequently,
there is a critical need to filter out unnecessary tables and columns,
directing the language models attention to relevant tables and columns with
schema-linking, to reduce errors during SQL generation. Previous approaches
have involved sorting tables and columns based on their relevance to the
question, selecting the top-ranked ones for sorting, or directly identifying
the necessary tables and columns for SQL generation. However, these methods
face challenges such as lengthy model training times, high consumption of
expensive GPT-4 tokens in few-shot prompts, or suboptimal performance in schema
linking. Therefore, we propose an inventive schema linking method in two steps:
Firstly, generate an initial SQL query by utilizing the complete database
schema. Subsequently, extract tables and columns from the initial SQL query to
create a concise schema. Using CodeLlama-34B, when comparing the schemas
obtained by mainstream methods with ours for SQL generation, our schema
performs optimally. Leveraging GPT4, our SQL generation method achieved results
that are comparable to mainstream Text-to-SQL methods on the Spider dataset.

使用 CodeLlama-34B 建立了一种创新的模式链接方法，通过从初始 SQL 查询中提取表格和列，创建了一个简明的架构，并且在与主流方法对比的情况下，在 SQL 生成方面表现最佳。利用 GPT4，在 Spider 数据集上实现了与主流 Text-to-SQL 方法相媲美的结果。

SQL-to-Schema 提升文本到 SQL 的模式链接

SQL-to-Schema Enhances Schema Linking in Text-to-SQL

Recent advancements in Text-to-SQL methods employing Large Language Models
(LLMs) have demonstrated remarkable performance. Nonetheless, these approaches
continue to encounter difficulties when handling extensive databases, intricate
user queries, and erroneous SQL results. To tackle these challenges, we present
\textbf{MAC-SQL}, a LLM-based multi-agent collaborative Text- to-SQL framework
based on LLMs. This framework comprises three agents: the \textit{Selector},
accountable for condensing voluminous databases and preserving relevant table
schemas for user questions; the \textit{Decomposer}, which disassembles complex
user questions into more straightforward sub-problems and resolves them
progressively; and the \textit{Refiner}, tasked with validating and refining
defective SQL queries. We perform thorough experiments on two Text-to-SQL
datasets, BIRD and Spider, attaining a state-of-the-art execution accuracy of
59.59\% on the BIRD test set. Moreover, we have open-sourced an instruction
fine-tuning model, \textbf{SQL-Llama}, based on Code Llama 7B, in addition to
an agent instruction dataset derived from training data based on BIRD and
Spider. The SQL-Llama model has demonstrated encouraging outcomes on the
development sets of both BIRD and Spider. However, when compared to the GPT-4
model, there remains a notable potential for enhancement. Our code and data can
be accessed publicly at
\href{https://github.com/wbbeyourself/MAC-SQL}{this https URL}.

最近的文本到 SQL 方法的发展使用了大型语言模型 (LLMs)，取得了显著的性能表现。然而，这些方法在处理庞大的数据库、复杂的用户查询和错误的 SQL 结果时仍然面临困难。为了解决这些挑战，我们提出了基于 LLMs 的多智能体协作文本到 SQL 框架 MAC-SQL。该框架由三个智能体组成：Selector 负责压缩庞大的数据库并保留用户问题的相关表模式；Decomposer 将复杂的用户问题分解为更简单的子问题并逐步解决；Refiner 负责验证和完善有缺陷的 SQL 查询。我们在两个 Text-to-SQL 数据集 BIRD 和 Spider 上进行了大量实验，在 BIRD 测试集上实现了 59.59% 的最先进的执行准确率。此外，我们还开源了基于 Code Llama 7B 的指导微调模型 SQL-Llama，以及基于 BIRD 和 Spider 训练数据的代理指令数据集。SQL-Llama 模型在 BIRD 和 Spider 的开发集上展示了令人鼓舞的结果，然而与 GPT-4 模型相比，仍有显著的提升潜力。我们的代码和数据可在 https://github.com/wbbeyourself/MAC-SQL 公开访问。