An Introduction to ULDBs and the Trio System

Abstract

We introduce ULDBs: relational databases that add uncertainty and lineage of the data as first-class concepts. The ULDB model underlies the Trio system under development at Stanford. We describe the ULDB model, then present TriQL, our SQL-based query language for ULDBs. TriQL's semantics over ULDBs is defined both formally and operationally, and TriQL extends SQL with constructs for querying lineage and confidence values. We also briefly describe our initial prototype Trio implementation, which encodes ULDBs in conventional relations and automatically translates TriQL queries into SQL commands over the encoding. We conclude with research directions for ULDBs and the Trio system.