Version 2 of the Knora API aims to make both the response and request formats more generic and consistent. Version 1 was basically the result of the reimplementation of the existing API of the SALSAH prototype. Since the development of this prototype has a long history and the specification of API V1 was an evolving process, V1 has various inconsistencies and peculiarities. With V2, we would like to offer a format that is consistent and hence easier to use for a client.
Please note that V2 is still in development. We do not yet recommend using it on productive systems.
Every request to API v2 includes
v2 as a path segment, e.g.
http://host/v2/resources/http%3A%2F%2Frdfh.ch%2Fc5058f3a. Accordingly, requests using any other version of the API will require another path segment.
Our preferred format for data exchange is JSON-LD. JSON-LD allows the Knora API server to provide responses that are relatively easy for automated processes to interpret, since their structure and semantics is explicitly defined. For example, each user-created Knora resource property is identified by an IRI, which can be dereferenced to get more information about it (e.g. its label in different languages). Moreover, each value has a type represented by an IRI. These are either standard RDF types (e.g. XSD datatypes) or more complex types whose IRIs can be dereferenced to get more information about their structure.
At the same time, JSON-LD responses are relatively easy for software developers to work with, and are more concise and easier to read than the equivalent XML. Items in a response can have human-readable names, which can nevertheless be expanded to full IRIs. Also, while a format such as Turtle just provides a set of RDF triples, an equivalent JSON-LD response can explicitly provide data in a hierarchical structure, with objects nested inside other objects.
Resources and entities are identified by IRIs. The format of these IRIs is explained in Knora IRIs.
Knora API v2 uses RDF data structures that are simpler than the ones actually stored in the triplestore, and more suitable for the development of client software. Thus we refer to the internal schema of data as it is stored in the triplestore, and to external schemas which are used to represent that data in API v2.
Knora API v2 offers a complex schema and a simple one. The main difference is that the complex schema exposes the complexity of value objects, while the simple version does not. A client that needs to edit values must use the complex schema in order to obtain the IRI of each value. A client that reads but does not update data can use the simplified schema. The simple schema is mainly intended to facilitate interoperability with other RDF-based systems in the context of Linked Open Data. It is therefore designed to use the simplest possible datatypes and to require minimal knowledge of Knora.
In either case, the client deals only with data whose structure and semantics are defined by external Knora API ontologies, which are distinct from the internal ontologies that are used to store date in the triplestore. The Knora API server automatically converts back and forth between these internal and external representations. This approach encapsulates the internals and adds a layer of abstraction to them.
IRIs representing ontologies and ontology entities are different in different schemas; see Knora IRIs.
Some API operations inherently require the client to accept responses in the complex schema. For example, if an ontology is requested using an IRI indicating the simple schema, the ontology will be returned in the simple schema (see Querying, Creating, and Updating Ontologies).
Other API operations can return data in either schema. In this case, the complex schema is used by default in the response, unless the request specifically asks for the simple schema. The client can specify the desired schema by using an HTTP header or a URL parameter:
- the HTTP header
- the URL parameter
Both the HTTP header and the URL parameter accept the values