Technology Category
Machine Learning: Large Multimodal Models (LMMs)Data Mining & Knowledge Management: Conversational Systems for Recommendation & RetrievalNatural Language Processing: (Large) Language Models
A Survey of LLM $ imes$ DATA
🤖 AI Summary
This paper addresses the insufficient bidirectional synergy between large language models (LLMs) and data management (DATA). To bridge this gap, it proposes a dual-path integration framework: DATA4LLM—establishing a data supply infrastructure for the full LLM lifecycle, incorporating techniques such as deduplication, synthetic data augmentation, KV cache optimization, RAG post-processing, and retrieval-augmented prompting; and LLM4DATA—leveraging LLMs as general-purpose data engines for novel paradigms in data manipulation, analysis, and system optimization. The work introduces the first systematic taxonomy spanning both LLM and database research domains, identifying 12 categories of technical challenges and surveying 87 state-of-the-art studies. This taxonomy provides a foundational theoretical framework and practical guidelines for designing AI-native data systems.
Technology Category
Application Category
📝 Abstract
The integration of large language model (LLM) and data management (DATA) is rapidly redefining both domains. In this survey, we comprehensively review the bidirectional relationships. On the one hand, DATA4LLM, spanning large-scale data processing, storage, and serving, feeds LLMs with high quality, diversity, and timeliness of data required for stages like pre-training, post-training, retrieval-augmented generation, and agentic workflows: (i) Data processing for LLMs includes scalable acquisition, deduplication, filtering, selection, domain mixing, and synthetic augmentation; (ii) Data Storage for LLMs focuses on efficient data and model formats, distributed and heterogeneous storage hierarchies, KV-cache management, and fault-tolerant checkpointing; (iii) Data serving for LLMs tackles challenges in RAG (e.g., knowledge post-processing), LLM inference (e.g., prompt compression, data provenance), and training strategies (e.g., data packing and shuffling). On the other hand, in LLM4DATA, LLMs are emerging as general-purpose engines for data management. We review recent advances in (i) data manipulation, including automatic data cleaning, integration, discovery; (ii) data analysis, covering reasoning over structured, semi-structured, and unstructured data, and (iii) system optimization (e.g., configuration tuning, query rewriting, anomaly diagnosis), powered by LLM techniques like retrieval-augmented prompting, task-specialized fine-tuning, and multi-agent collaboration.
Problem
Research questions and friction points this paper is trying to address.
Explores bidirectional integration of LLMs and data management systems
Examines data processing, storage, serving challenges for LLM workflows
Investigates LLM applications in data manipulation, analysis, and system optimization
Innovation
Methods, ideas, or system contributions that make the work stand out.
Data processing for scalable LLM training
Efficient storage hierarchies for LLMs
LLMs as engines for data management
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
No related jobs found.
