Where Does Metadata End, and the Graph Begin?

Metadata schemata are either general-purpose, or focused on domain- or format-specific description. It is not very common to find true hybrid schemata. Some schema provide profiles that accommodate additional format- or domain-specific elements - often as extended keywords.

This presents a problem in respect of mapping between schemata, since no single schema is able to accommodate all the elements from all schemata - especially of they are domain- or format-specific.

One solution to this problem is to ‘ignore’ the elements that are difficult to map, by including them in a target schema as ‘unmapped’ snippets with a meaningful label. In this way, the schematic aspect is not explicitly solved but the text snippet remains available for indexing against a controlled vocabulary.

Conceptually, this solution is similar to the one provided by NOSQL databases: allowing flexible and non-schematic extensions of arbitrary data structures to be stored as large text-based snippets. These can be indexed, preferably using an external vocabulary matching the original schema requirements, and can be processed for lexical relationships.

There are a set of common elements across most schemata that can be thought of as ‘core’ or ‘fundamental’. These elements are broadly similar to the elements contained in Dublin Core and DataCite schema.