The RDF Working Group held its first face-to-face meeting on 13 & 14 April in Amsterdam. After two months of elaborate discussions the WG used this meeting to focus on the task it is chartered to do, namely updating RDF to meet the established practice of RDF usage. Interested readers can find detailed minutes of the discussions on the meeting page. A synopsis of the highlights:
- The WG expects to publish a First Public Working Draft of a Turtle recommendation in about two months. We resolved several issues w.r.t syntactic details of Turtle. The Turtle document is likely to include a separate section/appendix on N-Triples as a limited subset of Turtle. N-Triples is typically used as a dump format for triples. We also discussed a quadruple extension of Turtle (with a separate media type), but we refrained from deciding on syntactic details until the issues mentioned under the next item have been resolved.
- The WG concluded that the meaning of the term “graph” is overloaded. For the moment the WG has identified three different notions of “graph” (see the graph terminology page) for which we still have to find adequate labels. The current terms, “g-box” (a graph container, the contents of which may change over time), “g-snap” (a snapshot of a g-box at some point in time) and “g-text” (a serialization of a graph in, for example, Turtle or RDF/XML) are just placeholders. We discussed which notions of graphs could/should have identifiers. We discussed intensively how the three graph notions relate to the use of the term graph in the SPARQL specifications (in general, consistency with SPARQL is a major concern of the WG). We expect this to lead to a revised version of the 2004 RDF Concepts document. We also discussed implications for the RDF Semantics document which we hope to be minimal (maybe even zero).
- Publication of a JSON format for RDF gave rise to large amounts of email traffic in the first two months of the WG. We identified three use cases for a JSON-RDF syntax.
The first use case is exemplified by the Talis JSON format. This JSON format mimics the RDF graph structure in the JSON syntax. This format is mainly useful for developers already familiar with RDF. The WG decided that it is useful and feasible to publish such a format as a W3C recommendation, taking the Talis proposal as starting point.
The third use case is the most challenging one: providing a simple JSON object-format which general Web developers can use to expose their data and which subsequently can be consumed by RDF tools. JSON-LD is an example of this approach. The idea is that Web developers do not need to be aware of the RDF triple model when exposing their data. Such a JSON format could have a very high impact. Unfortunately, we do not seem to have enough insight yet into the issues involved to standardize this or define best-practices for it. For the moment, we want to treat this as a high-potential incubator activity and plan to revisit the situation at a later stage of the WG life time (and hopefully publish something by that time).
In addition, we discussed a number of “clean-up” issues. The WG is likely to mark some parts of the 2004 RDF vocabulary as “archaic” (meaning: we have no intention of removing these, but advise refraining from usage in the future), but details of this still need to be agreed on at a later stage.
As chair I was in particular happy with the open and constructive atmosphere in which the discussions took place. The results of the WG are not likely to shake the Web world, but will hopefully contribute significantly to enhanced and widespread effective use of Semantic Web technology.
The RDF Working Group has a public comment list. We welcome any feedback from the community.