Text/Relational Database Management Systems: Overview and Proposed SQL Extensions

Combined text and relational database support is increasingly recognized as an emerging need of industry, spanning applications requiring text fields as parts of their data (e.g., for customer support) to those augmenting primary text resources by conventional relational data (e.g., for publication control). In this paper, we propose extensions to SQL2 that provide flexible and efficient access to structured text described by SGML or other encodings. We also propose an architecture to support a text/relational database management system as a federated database environment, where component databases are accessed via "agents": SQL agents that translate standard or extended SQL2 queries into vendor-specific dialects, and text agents that process text sub-queries on full-text search engines. Also available via anonymous ftp from site cs-archive.uwaterloo.ca, directory cs-archive/CS-95-25, file CS-95-25.ps.Z; and via the World Wide Web from page http://bluebox.uwaterloo.ca/OED/trdbms.htm..