A Variational Database Management System
summary
Framework encodes any structural or content variation using a Boolean feature space and stores all variants in one Haskell-built VDBMS layer over an RDBMS, tagging rows with a presence condition so each tuple keeps which variant(s) it belongs to. It extends relational algebra to variational relational algebra (choices, feature-tagged attributes, empty relation), typechecks queries with variation-preservation and proves soundness in Coq. SQL generators produce executable SQL; experimental comparisons of NBF/UBF/UAV strategies show tradeoffs with UAV and optimized variants often outperforming naive encodings.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:54Background Relational databases
- 2:03Motivation Examples of repeated pattern
- 3:43Databases used in software development
- 4:37Databases used in software analysis
- 6:25Background Variation space and its encoding
- 8:24Framework | A variational database
- 9:33Framework | Database variants
- 9:52Framework Variational relational algebra (VRA)
- 11:24Framework | Variation-preserving type system
- 12:55Implementation | VDBMS architecture
- 14:20Implementation Comparison of SQL generators
related talks
13mVBASE: Unifying Online Vector Similarity Search and Relational Queries via Relaxed...
20mGenSQL: A Probabilistic Programming System for Querying Generative Models of Database(…)
41mGetting Started With Vector Databases
45mA gentle introduction to Vector Databases
01h 01mBrief Tutorial on Probabilistic Databases
57mAlgebraic Databases
54mDissimilarity search...
12mVector Databases: Why QA professionals Needs to Care About them in the Age of AI?