Understanding Queries by Conditional Instances
summary
Defines conditional instances—small representative instances with global conditions—that capture distinct ways a domain-relational-calculus query can be satisfied; proves existence is undecidable in general and NP-hard subproblems are handled with a chase-inspired heuristic plus SMT solvers and Boolean how‑provenance for minimization. Produces a web UI and test-suite tooling used by 1,600 students over five years; a pilot user study finds concrete ground examples are easier to read but sets of conditional instances reveal multiple errors. Aggregates/HAVING remain unsupported.
AI-written summary · about
chapters (18)
- 0:00Database education tools
- 1:03Database course demand
- 2:10Debugging student SQL
- 3:21Autograder limitations
- 4:26Boolean how-provenance
- 6:38Motivation for conditional instances
- 8:31Conditional instances defined
- 10:01Query tree analysis Q2-Q1
- 11:05Requirements for solutions
- 12:35Enumerating homomorphisms
- 13:43Problem undecidability
- 14:45Chase procedure
- 16:30Drinker query example
- 19:07Handling quantifiers and disjunctions
- 21:13Query debugging user study
- 23:00Summary of contributions
- 24:26Open challenges
- 25:29Q&A on aggregates and LLMs
related talks
01h 10mQuery Optimization and Evaluation 4: Instance Optimality
38mThe Intensional-Extensional Problem in Probabilistic Databases
53mDirect Access for Conjunctive Queries with Negation
43mA Dichotomy for Queries with Negation in Probabilistic Databases
19mAccessing Answers to Unions of Conjunctive Queries with Ideal Time Guarantees: II
33mCombined Approximations for Probabilistic Query Evaluation: An Intensional Approach
01h 09mAccessing Answers to Unions of Conjunctive Queries with Ideal Time Guarantees: I
01h 01mBrief Tutorial on Probabilistic Databases
