FDO Profiles - Model Attributes

[juliancervos]

Central to the prototype Digital Twins (pDTs) are the models that power them. BioDTModel is one of the main types of digital objects of the project, and therefore it has its own FDO Profile with certain type-specific metadata attributes that are shown in the table below. Note that apart from those type-specific attributes, any FDO Model also has the general kernel attributes (discussed on #3).

In the following table, a set of attributes has been listed. As a starting point, we have drawn inspiration from the CodeMeta Project for the first few attributes. The goal was to select the minimum metadata attributes that any model should have in BioDT to start developing metadata descriptions of all models. As the project progresses and more attributes are needed, they can be incorporated or modified as we see fit.

Model-specific Attributes Table

Table format: the name of the metadata attributes is shown in the "Property" column, including a link to Schema.org (or another metadata schema from which the property was taken), and a definition in the "Description" column. The "Type" column specifies the expected values for each property, while the "Cardinality" column specifies the amount and whether they are optional or mandatory. Lastly, additional remarks might be added to the "Comments" column, and an example is given in the final column.

Property	Description	Type	Cardinality	Comments	Example
codeRepository	Link to the repository or repositories where the un-compiled, human-readable code is located.	URL	1/many	A link to GitHub, GitLab, or similar. Preferably, the code repository should be under the BioDT GitHub organisation so people in the project have access to it.	"codeRepository": "https://github.com/BioDT/biodt-fair"
contributor	Additional people or organisations that contributed to developing the model, despite not having authorship.	Organization Person	0/many	If possible, use an ORCiD (or ROR ID, for organizations), otherwise use full name in natural order.	"contributor": "https://orcid.org/0000-0003-0791-7164"
softwareVersion	The version of the model instance.	Text	1/1	It is advised to follow the semantic versioning guidelines.	"softwareVersion": "2.0.13"
programmingLanguage	The computer programming language the code is written in.	ComputerLanguage Text	1/many		"programmingLanguage": "Python"
softwareRequirements	Component dependency requirements for the model.	Text URL	0/many	This includes runtime environments and shared libraries that are not included as part of the model but are required to run it.	"softwareRequirements" : "https://pypi.org/project/cwltool/"
supportingData	Related datasets used for the model, for example, as training or input data.	Dataset DataFeed	1/many		"supportingData": {"@id": "https://doi.org/10.15468/dl.b3n3r9"}

The attributes from the previous table (together with the Kernel Attributes) are also specified as an [RO-Crate "template"], and a proper RO-Crate profile is in the works.

In this discussion post, we can also discuss different aspects of the FDO Model attributes. For example:

• Is there any aspect of the models used in BioDT that is currently missing and we should capture as metadata?