AF:III:Small:Collaborative Research: New Frontiers in Join Algorithms: Optimality, Noise, and Richer Languages

AF:III:Small:协作研究：连接算法的新领域：最优性、噪声和更丰富的语言

• 批准号: 1319402
• 负责人: Atri Rudra
• 金额: $ 32.61万
• 依托单位: SUNY at Buffalo
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-09-01 至 2017-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1319402&HistoricalAwards=false
• 关键词: AF; III; Small; Collaborative; Research

The relational join is central to relational database processing, which is the dominant way data is processed today. The join also models problems in biological and social networks, coding theory, compressed sensing, machine learning, and constraint satisfaction. Recently,  the investigators described the first ever worst-case optimal algorithm (the NPRR algorithm) for join queries.  These new results open a line of new tools to attack a diverse set of fundamental problems related to the join. This project aims to further exploit the new algorithmic techniques developed for NPRR to address the following three classes of problems:(1) Optimal Join algorithms. Developing algorithms that are instance optimal when the data are stored in either traditional database indexes or new indexing structures is a goal of this project. (2) Coping with and Leveraging Noise. This project will extend the latest work to handle and leverage both worst-case and statistical noise models, bridging to coding theory and compressed sensing.  (3) Expressive Query Languages. The project will explore a series of extensions to join queries that will pave the way to overcome challenges in motif finding, search, databases with functional dependencies, and more powerful classes of queries and join operations.If successful, the results of this grant will apply to a variety of pattern extraction problems in modern massive, dynamic, and noisy data sets, which have a wide range of applications in complex network analysis, coding theory, and compressive sensing.

关系连接是关系数据库处理的核心，是当今数据处理的主要方式。连接还对生物和社交网络、编码理论、压缩感知、机器学习和约束满足中的问题进行了建模。首个用于连接查询的最坏情况最优算法（NPRR 算法）​这些新结果开辟了一系列新工具来解决与连接相关的各种基本问题。 NPRR 解决以下三类问题：（1）开发当数据存储在传统数据库索引或新索引结构中时实例最优的算法是该项目的目标。该项目将扩展最新的工作来处理和利用最坏情况和统计​​噪声模型，连接编码理论和压缩感知。 (3) 表达查询语言该项目将探索一系列扩展来连接查询，从而为查询奠定基础。克服主题查找、搜索、具有功能依赖性的数据库以及更强大的查询和连接操作类方面的挑战的方法。如果成功，这笔资助的结果将适用于现代大规模、动态和噪声中的各种模式提取问题数据集，在复杂网络分析、编码理论和压缩感知方面有广泛的应用。