一句话：首次系统研究数据集成中的“抽象”问题：给定源模式上的查询，求其在全局模式上、模映射等价地刻画原查询的抽象查询，并给出形式化框架与研究展望。
数据集成系统通过全局模式、源模式以及二者之间的映射，为多个已有数据源提供统一而抽象的视图。过去数十年的研究主要关注反向问题：在全局模式上表达查询，将其改写为源上的查询并求值以得到答案。本文提出并综述一个全新的问题方向——抽象（abstraction）：给定一个直接表达在源上的查询，目标是找到它在全局模式上的对应查询，使其在映射的意义下刻画原查询。论文给出该问题的形式化框架与基本定义，梳理近期文献中关于抽象查询存在性、构造与语义刻画的主要结果，并讨论若干值得深入的研究方向。
Data integration provides a unified and abstract view over a set of existing data sources. The typical architecture of a data integration system comprises the global schema, which is the structure for the unified view, the source schema, and the mapping, which is a formal account of how data at the sources relate to the global view. Most of the research work on data integration in the last decades deals with the problem of processing a query expressed on the global schema by computing a suitable query over the sources, and then evaluating the latter in order to derive the answers to the original query. Here, we address a novel issue in data integration: starting from a query expressed over the sources, the goal is to find an abstraction of such query, i.e., a query over the global schema that captures the original query, modulo the mapping. The goal of the paper is to provide an overview of the notion of abstraction in data integration, by presenting a formal framework, illustrating the results that have appeared in the recent literature, and discussing interesting directions for future research.
DOI 原文 ·
@inproceedings{paperbot6162,
title = {Abstraction in Data Integration},
author = {Gianluca Cima and Marco Console and Maurizio Lenzerini and Antonella Poggi},
booktitle = {LICS},
year = {2021},
doi = {10.1109/lics52264.2021.9470716}
}