Developed as part of the CAMH knowledge graph project
A KO is a digital package containing one of more computable readable files and/or executable functions along with other information. Here we will focus on KOs containing individual pure functions. Each pure function has inputs (a.k.a. parameters) and outputs (a.k.a. computed results). Each pure function also has one or more concrete implementations. Clients execute a function by sending inputs/parameters and receiving outputs/computed results.
Initially, the KGrid used only functions with implementations intended to be exposed as a RESTful API. Each function was described by an OpenAPI doc (we called them endpoints) and mapped to an implementation in YAML for deployment in a particulat runtime environment (via deployment.yaml).
The Function Ontology (FNO) is capable of describing functions at a higher level and is particularly suited for knowledge engineering. The Provenance Ontology provides a similar capability when describing processes in which agents perform activities which transform entities. Both ontologies are useful in modeling the behavior and results of knowledge objects deployed in the Knowledge Grid or elsewhere.
This repo contains a sample KO with an FNO representation of its functions and results, and instructions for using it in one or more RDF/graph systems.
This sample uses a JSON-LD document in flattened form (a top-level anonymous node containing an @context element and an @graph element whose value is an array of nodes.) We think it will work with an expanded or compacted forms but have not tested.
The @context element is something like this:
{
"@context": {
"dc": "http://purl.org/dc/elements/1.1/",
"fno": "https://w3id.org/function/ontology#",
"koio": "http://kgrid.org/koio#",
"xsd": "http://www.w3.org/2001/XMLSchema#",
"sdo": "http://www.schema.org/",
"prov": "http://www.w3.org/ns/prov#",
"rdfs": "http://www.w3.org/2000/01/rdf-schema#"
},
"@graph": [{...}, ...]
}
The "@graph" elements defines the nodes in this sub-graph. It has this shape:
{
"@context": {...},
"@graph": [
{ "@type": "fno:Function", ...},
{ "@type": "fno:Execution", ...},
{ "@type": "fno:Output", ...},
{ "@type": "fno:Implementation", ...},
{ "@type": "fno:Mapping", ...}
]
}Adding the nodes represented by the JSON-LD document in this KO is somewhat platform dependent. Conult your platform documentation. Here's what worked for us in the following systems.