Relation-collapse: an optimisation technique for the similarity algebra SA

  • Authors:
  • Thomas Herstel;Ingo Schmitt

  • Affiliations:
  • Fakultät für Informatik, Universität Magdeburg, Magdeburg;Fakultät für Informatik, Universität Magdeburg, Magdeburg

  • Venue:
  • ADBIS'05 Proceedings of the 9th East European conference on Advances in Databases and Information Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Query systems of multimedia database systems should support similarity queries as well as user preferences like query term weighting. The graphical query language WS-QBE integrates these concepts and is a user-friendly query language. For evaluation purposes WS-QBE queries are translated into similarity algebra $\mathcal{SA}$ expressions. Expressions produced by the generation algorithm are very complex and thus need simplification and optimisation. One technique aiming at expression simplification is the relation-collapse technique. This technique, which is focus of this work, drastically reduces the number of basis relation scans and thus promises a more efficient query evaluation. Further, we discuss employing special, efficient implementations for algebra operations.