Tagger: A Tool for the Discovery of Tagged Unions in JSON Schema Extraction

Abstract

This tool demo features an original approach to model inference or schema extraction from collections of JSON documents: We automatically detect tagged unions, an established design pattern in hand-crafted schemas for conditionally declaring subtypes. Our “Tagger” approach is based on the discovery of conditional functional dependencies in a relational encoding of JSON objects. We have integrated our prototype implementation in an open source tool for managing data models in schema-flexible NoSQL data stores. Demo participants can interactively apply different schema extraction algorithms to real-world inputs, and compare the extracted schemas with those produced by “Tagger”.