NoSym: Non-Symbolic Databases for Data Decoupling

Under the Unique Name Assumption (UNA), users need to have shared agreements on signifiers to use in schema or data, e.g. to use “genre” and not “type” to refer to a movie’s category. Agreements are difficult in open environments such as datasets on the web, open data, and crowd-sourced databases, thus this assumption can be invalid. Schema matching and data integration can be limited in responding to this problem [2] as: (1) schemas might not be available a priori with schema-less data sources and queries becoming more common; (2) dataset-level schema/data mappings limit a user’s ability to provide a contextual interpretation of a signifier suitable for a specific query-data matching task; and (3) data integration typically has an overhead which hinders the availability and low latency of databases.