Scholarlink

In today’s data-driven world, efficient access to databases is paramount for various applications across industries. A significant change has been observed in human-computer interaction in the contemporary era of artificial intelligence. The nexus between database administration and natural language processing is where this shift is most noticeable. Existing methods of working with SQL databases require a high degree of technical proficiency, which limits the accessibility of these tools to non-technical users. Key objectives of this study included investigating the utilization of Large Language Models (LLM) to enhance data retrieval from SQL databases, focusing on user-friendly access and robust data security. Leveraging large language models, particularly pre-trained models, this research aimed to develop an end-to-end solution for database access, enabling users to interact with databases through natural language queries. The accuracy level was determined to be 82.6%. This paper focuses on data security in database access, highlighting the protection of sensitive information during retrieval. Experiments with real-world datasets showed improved user experience, query accuracy, and data security, validating the effectiveness of the AI-powered approach. The research findings have implications for various domains, including data management, information retrieval, and cybersecurity, paving the way for more efficient and secure database interaction in the era of big data and LLM.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging Large Language Models for Text-to-SQL with Attention to Data Security

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Jubayer Abdullah Joy,
  • Zahid Akon,
  • Partho Sarathi Sarker,
  • Md. Mahadi Hasan Shaon,
  • Md. Abdur Razzak,
  • Shanto Rahman,
  • Md. Erfan,
  • Sadib Hassan Rumman

摘要

In today’s data-driven world, efficient access to databases is paramount for various applications across industries. A significant change has been observed in human-computer interaction in the contemporary era of artificial intelligence. The nexus between database administration and natural language processing is where this shift is most noticeable. Existing methods of working with SQL databases require a high degree of technical proficiency, which limits the accessibility of these tools to non-technical users. Key objectives of this study included investigating the utilization of Large Language Models (LLM) to enhance data retrieval from SQL databases, focusing on user-friendly access and robust data security. Leveraging large language models, particularly pre-trained models, this research aimed to develop an end-to-end solution for database access, enabling users to interact with databases through natural language queries. The accuracy level was determined to be 82.6%. This paper focuses on data security in database access, highlighting the protection of sensitive information during retrieval. Experiments with real-world datasets showed improved user experience, query accuracy, and data security, validating the effectiveness of the AI-powered approach. The research findings have implications for various domains, including data management, information retrieval, and cybersecurity, paving the way for more efficient and secure database interaction in the era of big data and LLM.