Functional Query Optimization over Object-Oriented Views for Data Integration

  • Authors:
  • Vanja Josifovski;Tore Risch

  • Affiliations:
  • Laboratory for Engineering Databases, Linköping University, 58183 Linköping, Sweden. vanja@ida.liu.se;Laboratory for Engineering Databases, Linköping University, 58183 Linköping, Sweden. torri@ida.liu.se

  • Venue:
  • Journal of Intelligent Information Systems - Special issue on functional approach to intelligent information systems
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

AMOS is a mediator system that supports passive (non-intrusive)integration of data from heterogeneous and autonomous data sources. Itis based on a functional data model and a declarative functional querylanguage AMOSQL. Foreign data sources, e.g., relational databases, textfiles, or other types of data sources can be wrapped with AMOSmediators, making them accessible through AMOSQL. AMOS mediators cancommunicate among each other through the multi-database constructs ofAMOSQL that allow definition of functional queries and OO viewsaccessing other AMOS servers. The integrated views can contain bothfunctions and types derived from the data sources. Furthermore, localdata associated with these view definitions may be stored in themediator database. This paper describes AMOS‘ multi-database queryfacilities and their optimization techniques. Calculus-based functiontransformations are used to generate minimal query expressions beforethe query decomposition and cost-based algebraic optimization stepstake place. Object identifier (OID) generation is used forcorrectly representing derived objects in the mediators. A selectiveOID generation mechanism avoids overhead by generating in themediator OIDs only for those derived objects that are either neededduring the processing of a query or have associated local data in themediator database. The validity of the derived objects that areassigned OIDs and the completeness of queries to the views areguaranteed by system generated predicates added to the queries.