Wrapping VRXQuery with self-adaptive fuzzy capabilities

  • Authors:
  • Keivan Kianmehr;Tansel Ö/zyer;Mehmet Kaya;Reda Alhajj

  • Affiliations:
  • Department of Computer Science, University of Calgary, Calgary, Alberta, Canada;Department of Computer Engineering, TOBB University, Ankara, Turkey;Department of Computer Engineering, Furat University, Elazı/ğ/, Turkey;(Correspd. alhajj@cpsc.ucalgary.ca) Department of Computer Science, University of Calgary, Calgary, Alberta, Canada/ he is also affiliated with Department of Computer Science, Global University, B ...

  • Venue:
  • Web Intelligence and Agent Systems
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper addresses the development of a plug and run wrapper to incorporate fuzziness into VRXQuery, the querying facility of VIREX which is a user-friendly system for transforming and querying relational data as XML. Our basic argument is not to force the underlying XML data to incorporate fuzziness. Rather, fuzziness is smoothly supported in a novel plug and run manner via a wrapper. Either the user specifies the membership functions for the elements/attributes to be queried as fuzzy, or multi-objective genetic algorithm is used to automatically decide on and optimize the membership functions. As the database is dynamic and updates are very frequent, there is no guarantee that existing membership functions will continue to properly reflect the current database state; this is identified when the cluster homogeneity deviates by a threshold. For the former case, the user will receive a warning; and for the latter case, the system will automatically adjust the clustering and the membership functions. The interface of VIREX has been expanded to allow specifying queries with fuzziness. Then, queries expressed in VRXQuery empowered with fuzziness are translated into corresponding XQuery code, which is run on the underlying XML and the returned result is translated into a fuzzy representation; translation into SQL is also possible. The user is given the choice to display the result either as colored-text or in graphical format.