Content-Length: 8381 | pFad | http://www.w3.org/submissions/2024/01/Comment/

| Submissions

Team Comment on "The PROV-JSONLD Serialization. A JSON-LD Representation for the PROV Data Model" Submission

W3C is pleased to receive the “The PROV-JSONLD Serialization” submission from King's College London.

Summary

The PROV family of specifications was published in 2013 by the Provenance Working Group. It provides a data model for provenance description [PROV-DM], which can be represented in various formats: a dedicated textual notation [PROV-N], an XML-based format [PROV-XML], or any RDF concrete syntax by leveraging the PROV ontology [PROV-O]. Another format, based on JSON, was also proposed as a previous member submission [PROV-JSON].

This member submission complements that family with a new JSON-based format: PROV-JSONLD. This format leverages the capabilities of JSON-LD 1.1 to fulfill four requirements identified by the authors, which none of the previous formats satisfy all at once: lightweight, natural, semantic and efficient.

The Submission is composed of four documents:

• The specification of the PROV-JSONLD format (in prose, with example code snippets)
• A JSON Schema capturing the syntactic constraints of that format
• A JSON-LD Context capturing the mapping of the formats with the PROV-O ontology
• A small extension of the PROV-O ontology, serialized in Turtle

Relationships to W3C Activities

Disclaimer: Publication of a Member Submission by W3C does not imply endorsement by W3C, including the W3C Team or Members, nor does it guarantee that a Working Group will agree to take any specific action on a Submission.

There are currently no W3C group maintaining the PROV family of specifications (neither Working Group, Community Group, Interest Group nor Business Group). Nevertheless, these specifications are still relevant to the Data ecosystem of W3C. They are still influencial in recently published specifications, such as the Data Catalog Vocabulary (DCAT) - Version 3.

Beyond its interest for the PROV family of specifications, this submission is also an examplary case of using JSON-LD to seemlessly bridge the gap between an easy-to-use JSON format and an underlying Linked Data conceptual model. This is achieved by providing a JSON schema and a JSON-LD concept, which were jointly designed, and leaving developers free to use one or the other (or both) depending on their preferred technology stack. This can be used as feedback for the JSON-LD Working Group for elliciting best practices in designing JSON-LD-based formats.

Next Steps

The PROV-JSONLD format is proposed as a Member Submission to serve as a stable resource for the Provenance community and the Semantic Web community. Ongoing maintenance is not anticipated at this time.

The following minor editorial changes are suggested before publication (but none of them are blockers):

• the foaf:mbox of the agent is the empty string in Example 1 and Section 6, while in Figure 1, it is <mailto:derek@example.org>. It is not clear if this is intended of if this should be fixed.
• In Section 4, 2nd paragraph: "JSON scheme" should be changed to "JSON schema".
• In Section 4.1: "String values" should be changed to "String values (lang_string)".
• In Section 4.1, last block of JSON-Schema ("ArrayOfValues"):
  • the indentation is not consistent,
  • the block also defines Context which is not mentioned in the text (and is actually described later in the section, see below).
• Section 4.19 is titled "prov:Bundle and prov:Document" but it also contains a description of contexts. It would seem more consistent to describe contexts in a separate section.
• In section 4.19, the definition of contexts says that the mapping should be from prefixes to URIs.
  • This should be "IRIs" instead of "URIs".
  • The JSON-Schema definition allows a mapping from any string to any string. Maybe it could be made more constraining, in accordance with the normative text?
• Sections 5.4, 5.10, 5.11, 5.12, 5.13 and 5.14 use the term "range" to denote the type coercion specified in the JSON-LD context. The term "range" might be mistaken for the concept of rdfs:range here, which is different from JSON-LD's type coercion. I suggest to change sentences of the form "The range of X is Y" to "The value of X is coerced to datatype Y."
• In Section 6, the definition of IC1: "despite being" should be changed to "despite it being" or to "even if".