데이터 마켓에 대한 포괄적인 조사
데이터 마켓은 데이터 상품의 가치를 극대화하기 위해 데이터 소유자, 구매자, 브로커 및 정책 입안자 등 다양한 주체가 상호 작용하는 복잡한 생태계입니다.
LSMGraph：一種採用多級 CSR 的高效能動態圖儲存系統
LSMGraph 是一種結合了 LSM-tree 的寫入效能和 CSR 的讀取效能優勢，專為高效處理動態圖資料而設計的新型儲存系統。
데이터베이스 무결성 검험 시 불일치성 허용 접근 방식: GIGO 문제에 대한 새로운 시각
기존 데이터베이스 무결성 검사 방식의 '완벽한 일관성' 요구사항을 완화하여, 불일치 데이터를 허용하면서도 무결성을 유지하는 새로운 접근 방식을 제시합니다.
Evaluating the Feasibility of Large Language Models as In-Context Databases with Dynamic Update Capabilities
This research paper explores the potential of large language models (LLMs) to function as dynamic in-context databases, capable of performing CRUD operations on data stored entirely within their context windows.
TrajRoute: A Trajectory-Based Routing System That Bypasses Traditional Map and Traffic Data
TrajRoute presents a novel routing paradigm that leverages raw historical trajectory data to compute efficient routes, potentially offering a lower-maintenance alternative to conventional map and traffic-based systems.
KVACCEL: A Host-SSD Collaborative Write Accelerator for Enhanced LSM-Tree Performance in Key-Value Stores
KVACCEL is a novel hardware-software framework that leverages a dual-interface SSD to eliminate write stalls in LSM-tree-based Key-Value Stores, improving throughput and resource utilization without requiring additional hardware costs.
DocETL: A System for Optimizing LLM-Based Complex Document Processing Pipelines
DocETL is a novel system designed to optimize complex document processing pipelines for accuracy by leveraging LLM agents to rewrite and evaluate user-defined pipelines, addressing the limitations of existing declarative frameworks that prioritize cost reduction over accuracy.
HyperBlocker: A GPU-Accelerated System for Efficient Rule-Based Blocking in Entity Resolution
HyperBlocker is a novel system that significantly accelerates rule-based blocking in Entity Resolution by leveraging GPUs, a pipelined architecture, and data-aware and rule-aware optimizations, outperforming both CPU-based and existing GPU-based solutions.
A Comprehensive Benchmarking Study Reveals Limitations in Synthetic Relational Data Generation
Current state-of-the-art methods for synthesizing relational data struggle to fully capture the complexity of real-world datasets, particularly in preserving multi-table relationships, impacting both data fidelity and utility for downstream machine learning tasks.
Enhancing Structured-Data Retrieval with GraphRAG: A Case Study on Soccer Data
Structured-GraphRAG, a versatile framework, enhances information retrieval across structured datasets by leveraging multiple knowledge graphs to provide more accurate and comprehensive responses to natural language queries.