Intelligent databases: Old challenges and new opportunities

摘要

The evolution of existing information systems and a new wave of data-intensive applications are creating a strong demand for database-centered programming environments much more sophisticated and intelligent than those supported by current database systems. In this paper, we describe the contributions that deductive databases offer to the evolution of databases and information systems to satisfy said demands. In addition to all database essentials, deductive databases support rule-based logic-oriented languages that allow terse formulations of complete applications, along with reasoning and queries. Thus, they support a rule-based interface that eliminates the impedance mismatch problem (between programming language and query sublanguage) and elevates the design and development of database applications to the level of declarative, knowledge-based specifications. In this paper, we review the evolution of the enabling technology and architectures of deductive database prototypes; then we focus on their applications, as seen by the author through his experience with theLDL/LDL++ project. In particular, the paper describes the languages and the (bottom-up) execution technology used by the first generation of deductive database prototypes. Then the paper discusses how the experience with a first-generation system (LDL) guided the design and implementation of a second-generation prototype (LDL++).