Abstract
Purchase PDF (475 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (6)
PDF (351 K)
doi:10.1016/j.is.2005.12.001 
Copyright © 2006 Elsevier B.V. All rights reserved.
Branch-and-bound processing of ranked queries
Received 22 April 2003;
References and further reading may be available for this article. To view references and further reading you must purchase this article.
Abstract
Despite the importance of ranked queries in numerous applications involving multi-criteria decision making, they are not efficiently supported by traditional database systems. In this paper, we propose a simple yet powerful technique for processing such queries based on multi-dimensional access methods and branch-and-bound search. The advantages of the proposed methodology are: (i) it is space efficient, requiring only a single index on the given relation (storing each tuple at most once), (ii) it achieves significant (i.e., orders of magnitude) performance gains with respect to the current state-of-the-art, (iii) it can efficiently handle data updates, and (iv) it is applicable to other important variations of ranked search (including the support for non-monotone preference functions), at no extra space overhead. We confirm the superiority of the proposed methods with a detailed experimental study.
Keywords: Databases; Ranked queries; R-tree; Branch-and-bound algorithms
Article Outline
- 1. Introduction
- 2. Related work
- 3. Problem definition
- 4. Ranked search on monotone preference functions
- 4.1. Problem characteristics and BRS
- 4.2. I/O optimality
- 4.3. Query cost estimation
- 4.4. Reducing the size of the R-tree
- 5. Alternative types of ranked queries
- 6. Experiments
- 6.1. Evaluation of conventional ranked search
- 6.1.1. Query cost comparison
- 6.1.2. Quality of cost prediction
- 6.1.3. Effect of space reduction
- 6.1.4. Performance for non-linear monotone functions
- 6.2. Evaluation of complex ranked search
- 7. Conclusion
- Acknowledgements
- References
