New techniques for studying set languages, bag languages and aggregate functions

  • Authors:

  • Leonid Libkin;Limsoon Wong

  • Affiliations:

  • Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA;Department of Computer and Information Science, University of Pennsylvania, Philadelphia, PA

  • Venue:
  • PODS '94 Proceedings of the thirteenth ACM SIGACT-SIGMOD-SIGART symposium on Principles of database systems
  • Year:
  • 1994

Quantified Score

Hi-index 0.00

Visualization

Abstract

We provide new techniques for the analysis of the expressive power of query languages for nested collections. These languages may use set or bag semantics and may be further complicated by the presence of aggregate functions. We exhibit certain classes of graphs and prove that the properties of these graphs that can be tested in such languages are either finite or cofinite. This result settles the conjectures of Grumbach, Milo, and Paredaens that parity test, transitive closure, and balanced binary tree test are not expressible in bag languages like the PTIME fragment of BALG of Grumbach and Milo and BQL of Libkin and Wong. Moreover, it implies that many recursive queries, including simple ones like the test for a chain, cannot be expressed in a nested relational language even when aggregate functions are available. In an attempt to generalize the finite-cofiniteness result, we study the bounded degree property which says that the number of distinct in- and out-degrees in the output of a graph query does not depend on the size of the input if the input is “simple”. We show that such a property implies a number of inexpressibility results in a uniform fashion. We then prove the bounded degree property for the nested relational language.