arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.02643cs.DBcs.AIcs.CLcs.LG
收藏

LLM增强的数据管理

LLM-Enhanced Data Management

Tsinghua University(清华大学)

Xuanhe Zhou, Xinyang Zhao, Guoliang Li

AI总结:

针对传统ML在数据管理中泛化与推理不足及LLM的幻觉、高成本、低准确率问题,提出LLMDB范式,通过微调、向量数据库和智能体实现高准确性、低成本与泛化性,支持查询重写、数据库诊断和数据分析。

摘要:

机器学习(ML)技术在优化数据管理问题方面已在近五年内得到广泛研究和部署。然而,传统ML方法在泛化性(适应不同场景)和推理能力(理解上下文)方面存在局限性。幸运的是,大型语言模型(LLMs)在理解上下文方面展现出高泛化性和与人类竞争的能力,这使其有望用于数据管理任务(如数据库诊断、数据库调优)。然而,现有LLMs存在若干局限:幻觉、高成本以及复杂任务准确性低。为应对这些挑战,我们设计了LLMDB,一种LLM增强的数据管理范式,它具备泛化性和高推理能力,同时避免幻觉、降低LLM成本并实现高准确性。LLMDB通过LLM微调和提示工程嵌入领域特定知识以避免幻觉。LLMDB通过向量数据库提供语义搜索和缓存能力来降低LLMs的高成本。LLMDB通过LLM智能体提供多轮推理和流水线执行来提高任务准确性。我们展示了LLMDB能良好支持的三个实际场景,包括查询重写、数据库诊断和数据分析。我们还总结了LLMDB的开放研究挑战。

英文摘要:

Machine learning (ML) techniques for optimizing data management problems have been extensively studied and widely deployed in recent five years. However traditional ML methods have limitations on generalizability (adapting to different scenarios) and inference ability (understanding the context). Fortunately, large language models (LLMs) have shown high generalizability and human-competitive abilities in understanding context, which are promising for data management tasks (e.g., database diagnosis, database tuning). However, existing LLMs have several limitations: hallucination, high cost, and low accuracy for complicated tasks. To address these challenges, we design LLMDB, an LLM-enhanced data management paradigm which has generalizability and high inference ability while avoiding hallucination, reducing LLM cost, and achieving high accuracy. LLMDB embeds domain-specific knowledge to avoid hallucination by LLM fine-tuning and prompt engineering. LLMDB reduces the high cost of LLMs by vector databases which provide semantic search and caching abilities. LLMDB improves the task accuracy by LLM agent which provides multiple-round inference and pipeline executions. We showcase three real-world scenarios that LLMDB can well support, including query rewrite, database diagnosis and data analytics. We also summarize the open research challenges of LLMDB.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑