Nonmaterialized relations and the support of information retrieval applications by relational database systems

✍ Scribed by Lynch, Clifford A.

• Publisher: John Wiley and Sons
• Year: 1991
• Tongue: English
• Weight: 946 KB
• Volume: 42
• Category: Article
• ISSN: 0002-8231
• DOI: 10.1002/(sici)1097-4571(199107)42:6<389::aid-asi1>3.0.co;2-l

✦ Synopsis

The exploratory, interactive searching process typically used in information retrieval applications is not wellsupported by current relational database management systems (RDBMSs). In particular, relational systems suffer serious performance problems. A new query processing strategy called nonmaterialized relations is proposed which should significantly improve the ability of relational systems to support IR applications without requiring changes to the SQL query language interface. The nonmaterialized relation strategy is a lazy evaluation method in which queries are evaluated only partially, and the rest of the computation is delayed until it is clear that it is required. The nonmaterialized relation evaluation technique also permits a multiuser database management system (DBMS) additional flexibility in balancing workload and response time as system loading varies over time.