User-Defined Aggregates in Database Languages

  • Authors:
  • Haixun Wang;Carlo Zaniolo

  • Affiliations:
  • -;-

  • Venue:
  • DBPL '99 Revised Papers from the 7th International Workshop on Database Programming Languages: Research Issues in Structured and Semistructured Database Programming
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

User-defined aggregates (UDAs) can be the linchpin of sophisticated data mining functions and other advanced database applications, but they find little support in current database systems. In this paper, we describe the SQL-AG prototype that overcomes these limitations by supporting UDAs as originally proposed in Postgres and SQL3. Then we extend the power and flexibility of UDAs by adding (i) early returns, (to express online aggregation) and (ii) syntactically recognizable monotonic UDAs that can be used in recursive queries to support applications, such as Bill of Materials (BoM) and greedy algorithms for graph optimization, that cannot be expressed under stratified aggregation. T his paper proposes a unified solution to both the theoretical and practical problems of UDAs, and demonstrates the power of UDAs in dealing with advanced database applications.