On the complexity of resilience for aggregation queries

Abstract

Resilience, as an potential explanation of a specified query, plays a fundamental and important role in query explanation, database debugging and error tracing. Resilience decision problem is defined on a database d, given a boolean query q where q(d) is initially true, and an integer k, it is to determine if there exists a tuple set Δ such that size of Δ is no more than k and query result q(d⊕Δ) becomes false, where ⊕ can be deletion or insertion operation. Results of this problem on relational algebraic queries have been showed in previous work. However, we revisit this decision problem on aggregation queries in the light of the parametric refinement of complexity theory, provide new results. We show that, this problem is intractable on nested COUNT and SUM query both under data complexity and parametric complexity..