Queries with Arithmetic on Incomplete Databases

摘要: The standard notion of query answering over incomplete database is that certain answers, guaranteeing correctness regardless how data interpreted. In majority real-life databases, relations have numerical columns and queries use arithmetic comparisons. Even though the answers still applies, we explain it becomes much more problematic in situations when missing occurs columns.We propose a new general framework allows us to assign measure certainty answers. We test agnostic scenario where do not prior information about values attributes, similarly predominant approach handling which assumes each null can be interpreted as an arbitrary value domain. key technical challenge lack uniform distribution entire domain such real numbers. overcome this by associating with asymptotic behavior volumes some subsets Euclidean space. show well-defined, describe approaches computing approximating it. While computationally hard, or result irrational number, even for simple constraints, produce polynomial-time randomized approximation schemes multiplicative guarantees conjunctive queries, additive first-order queries. also set experimental results confirm feasibility approach.