Data Modifications and Versioning in Trio

This paper presents the first DBMS for uncertain data that incorporates data modifications and a simple versioning system. Our work is in the context of Trio, a project at Stanford for managing data uncertainty and lineage. We establish SQL-based language constructs for data modifications, and an extended data model ULDB that supports these modifications yielding versioned relations. We show that Trio’s lineage feature enables answering a wide class of queries over ULDBs efficiently. We also give algorithms that leverage ULDB’s lineage for propagating data-modifications to Trio’s derived relations efficiently and incrementally. We have incorporated the data modification and versioning capabilities in the Trio system, and we validate our techniques through experiments.