[Paper Review] Counting Answers to Existential Positive Queries: A Complexity Classification

Hubie Chen, Stefan Mengel|arXiv (Cornell University)|Jan 13, 2016

Data Management and Algorithms13 references4 citations

TL;DR

This paper establishes a trichotomy theorem for the parameterized complexity of counting answers to existential positive queries on finite structures of bounded arity. By proving an equivalence theorem that reduces existential positive queries to primitive positive queries in terms of counting complexity, the authors unify and generalize prior results, showing that each problem is either fixed-parameter tractable, equivalent to the clique problem, or as hard as counting cliques—under standard complexity assumptions.

ABSTRACT

Existential positive formulas form a fragment of first-order logic that includes and is semantically equivalent to unions of conjunctive queries, one of the most important and well-studied classes of queries in database theory. We consider the complexity of counting the number of answers to existential positive formulas on finite structures and give a trichotomy theorem on query classes, in the setting of bounded arity. This theorem generalizes and unifies several known results on the complexity of conjunctive queries and unions of conjunctive queries.

Motivation & Objective

• To classify the parameterized complexity of counting satisfying assignments to existential positive queries on finite structures with bounded arity.
• To unify and generalize existing complexity results on conjunctive queries, unions of conjunctive queries, and related counting problems.
• To establish a theoretical foundation for understanding tractability in database query evaluation with counting, particularly in decision support systems.
• To extend prior dichotomy and trichotomy results on primitive positive queries to the broader class of existential positive queries.
• To provide a syntactic characterization of counting equivalence between primitive positive formulas, enabling complexity-preserving reductions.

Proposed method

• Introduce the concept of 'all-free part' of an existential positive formula, isolating its quantifier-free, free-variable components.
• Define a transformation that maps each existential positive formula to a set of primitive positive formulas, preserving counting complexity.
• Prove the 'equivalence theorem' (Theorem 3.1), showing that for every set of existential positive queries, there exists a corresponding set of primitive positive queries with identical counting complexity.
• Use a counting slice reduction to relate the counting problem on existential positive formulas to their transformed primitive positive counterparts.
• Leverage an existing trichotomy theorem for primitive positive queries as a black-box to derive the main trichotomy for existential positive queries.
• Apply syntactic conditions to determine when two primitive positive formulas are counting equivalent, generalizing classical work by Chandra and Merlin to the counting setting.

Experimental results

Research questions

• RQ1 What is the parameterized complexity of counting answers to existential positive queries on finite structures with bounded arity?
• RQ2 Can the complexity of counting answers to existential positive queries be reduced to that of primitive positive queries without loss of generality?
• RQ3 Under what syntactic conditions are two primitive positive formulas counting equivalent across all finite structures?
• RQ4 How does the presence of sentence disjuncts (i.e., quantifier-free sentences) affect the counting complexity of existential positive formulas?
• RQ5 Is there a uniform classification framework that unifies prior dichotomy and trichotomy results for conjunctive and union of conjunctive queries?

Key findings

• The paper establishes a trichotomy theorem: for any set of existential positive queries of bounded arity, the counting problem is either fixed-parameter tractable, equivalent to the clique problem, or as hard as counting cliques.
• The equivalence theorem (Theorem 3.1) shows that every set of existential positive queries is computationally equivalent in counting complexity to a corresponding set of primitive positive queries.
• Counting equivalence between primitive positive formulas is characterized syntactically: two formulas are counting equivalent if and only if they share the same set of minimal models and satisfy specific logical entailment conditions.
• The transformation from existential positive to primitive positive formulas preserves the number of satisfying assignments on every finite structure, enabling complexity transfer.
• The trichotomy result holds under the standard hypothesis that FPT ≠ W[1], implying that fixed-parameter tractability can be precisely characterized for this class.
• The assumption of bounded arity is only required for the final trichotomy, not for the equivalence theorem, suggesting potential extension to unbounded arity if a corresponding counting trichotomy for primitive positive queries were available.