CAREER: Extensible Personalization of Spatial and Spatio-temporal Database Management Systems

点击翻译按钮获取中文摘要

The goal of this research project is to allow database users to enjoy high quality personalized answers for their spatial and spatio-temporal queries, without sacrificing the query execution efficiency. As the concept of personalization is subjective, there are plethora of approaches that define the meaning and contents of a high quality personalized answer. This project does not aim to provide more definitions of personalization, instead, it provides a system-oriented approach that can accommodate most of the existing and future definitions in one highly efficient database engine. To achieve its goal, this project employs an extensible approach where core database modules for personalized queries are realized only once inside the database engine, and used many times in different ways to support a wide variety of personalized queries. The project provides such extensible core modules for:(a) single-table operations, e.g., selection and aggregations,(b) multi-table operations, e.g., join, (c) operations with adaptive cost that can only be computed online, and (d) handling uncertain data. With these extensible modules, registering a new personalization definition becomes as easy as plugging few modules in an easy-to-use interface. Once plugged in, the new personalized definition lives in the database engine, and enjoys the power of the extensible modules for query processor, optimizer, and indexing. Such extensible approach is highly attractive to industry as it gives minimal changes in the database engine to support wide variety of personalized queries, as well as orders of magnitude performance over existing approaches that are built on top of database systems. Besides its impact on industry, this project will have significant broader impact across multiple segments of society that include enhancing productivity, graduate and undergraduate student education, outreach to K-12 students, curriculum development, and tutorial presentations. Publications, technical reports, open-source software, and experimental data from this research are disseminated via the project web site (http://www.cs.umn.edu/~mokbel/career).