Abstract
While interest in the use of Prolog for database applications is growing, the size of such applications is limited by the ability of current implementations of Prolog for handling disc-resident clauses. Various software and hardware approaches, such as codeword indexing, have been put forward to solve this problem. The paper reports on a study of the comparative performances of one-level and two-level codeword indexing. Some benchmarks have been used to evaluate the two strategies. The performance figures obtained show that the two-level codeword indexing strategy provides a more efficient retrieval method in most cases, though it has drawbacks in some cases. To improve the performance of the codeword indexing strategy, an improved approach is presented that combines both one-level and two-level indexing. © 1991.
Original language | English |
---|---|
Pages (from-to) | 157-162 |
Number of pages | 6 |
Journal | Information and Software Technology |
Volume | 33 |
Issue number | 2 |
Publication status | Published - Mar 1991 |
Keywords
- codeword indexing
- database retrieval
- databases
- Prolog