Creating Schema Elements from Scratch
One main application of the Data Type Registry is the creation of elements of metadata schemas. Registering metadata elements has many advantages. These include the unambigious identification via a PID, additional metadata information, and the possibility of reusability and interoperability of types. In this tutorial, we will walk through the entire creation of schema elements, starting with the fundamental building blocks and ending at the generation of JSON schemas of complex data types.
What are Schema Elements in the DTR?
When clicking Create, the possible types that are available can seem overwhelming at first. That is why we want to shed some light on
the kinds of data types relevant for the creation of schema elements. In general, we differentiate between two different structures with
different responsibilities. BasicInfoTypes can be considered the basic building blocks, or leaf nodes, for schema elements. Each BasicInfoType is simply
some fundamental data type, with additional metadata and the possibility to apply some restrictions on that type. An InfoType on the other
hand is a more complex construct. It shares the set of metadata with BasicInfoTypes, but instead of a fundamental data type it represents a
composite of other types. It can either take the form of an Object, which is basically a set of key - value pairs where the values are other types,
or an Array, where one type can be repeated. The Tuple property will be implemented soon. The following table gives a quick overview over which type
can represent what:
| BasicInfoType | InfoType |
|---|---|
| String | Object |
| Number | Array |
| Boolean | Tuple |
We will start now by creating a BasicInfoType to descibe a single property of an existing metadata schema.
Creating a BasicInfoType
For the creation of our BasicInfoType we will use an actual metadata schema as a blueprint. This hopefully helps in following along the parallels and how
the schema is created hierarchically in the DTR. Take a look at the following schema from KonsortSWD, which develop infrastructure for research data for the
social sciences in the scope of the NFDI.
When untangling the $ref properties we see that the schema is basically an object containing just an array called variables, where each variable has a number of properties. To now recreate this schema in the DTR we have to start from the bottom and gradually build the schema up hierarchically. So let us zoom in on one of those
basic properties:
"title": {"type": "string", "minLength": 1, "maxLength": 550}
We see that it represents the fundamental data type string and applies restrictions relating to the minimum and maximum length of the string. Now we want to create this property as BasicInfoType in the DTR. Of course this is all possible via the API, but we consider the process via the UI for now.
Generating a JSON Schema
Lorem Ipsum