In light of discussion in #33, #37, #41, and #43, its clear that a lot of work remains regarding the co-existence of RDF content (the semantic web) and plain web content (html, xml, png) for the same resource. Linked data in the header of an HTML resource serves a different purpose and is used in a different context than linked data that is part of a graph of linked data generally. SELFIE is not going to be able to solve the riddle of unifying JSON-LD for the semantic web and JSON-LD for search engine and other crawler consumption.
There is a lot more work to do to unify these two (RDF content and HTML content with embedded linked data for the same URL). It's my opinion that acting as if we will be able to make recommendations in this space is naive. We can offer findings and make recommendations about some specific technologies in the context of an incremental progression.
The aim of ELFIE, as stated in our charter was:
The ELFIE is intended to test existing OGC and W3C standards with the goal of establishing a best practice for exposing links between and among environmental domain and sampling features in a highly adoptable standards compliant way that is compatible with modern web search technology.
and the am of SELFIE:
Building on the content-focused outcomes of the first ELFIE, the Second ELFIE (SELFIE) is designing and vetting Web-resource model and network behavior for cross-domain linked feature data that compliments and uses WFS3 as a building block. This aims to answer the question, how do we use linked data in a way that's compatible with W3C best practices and leverages OGC standards?
Given this, I want to propose that we focus further discussion in SELFIE as follows.
OGC-API Features and how it has been implemented should be a component of anything we recommend.
How to embed actionable links between and among feature URIs in existing OGC-API capabilities (html with linked data and plain geoJSON).
Additional media-types and APIs other than OGC-API features, while of great interest to the IE are too much for us to include in this work.
Each use case in the engineering report could and should describe the additional capabilities that were possible (or not) in addition to the SELFIE core.
How additional media-types (RDF-* in particular) play with the core OGC-API media types should be out of scope and we should discuss the difficulties we experienced with this area in the engineering report.
I'm running the risk of making this too long to matter, but one last thing to illustrate this proposal.
In reality, if an implementation were to have an API that could provide an html page or geoJSON data for a given URL, the html would have a <script> tag containing JSON-LD inspired by the SELFIE-index context. and the geoJSON would be just that. Any additional capabilities to present the linked data (a JSON-LD or RDF mimetype) or some other linked data for the url is essentially out of scope for SELFIE. While there may be some or even complete overlap between the linked data in the <script> tag, there is not necessarily a 1:1 correspondence between the two. To be very clear -- out of scope means we are not going to say we've found answers worth recommending as common practices in this space -- there are many ongoing projects that are working on those answers and we should all take part in them as part of the long term arc of development here.
In light of discussion in #33, #37, #41, and #43, its clear that a lot of work remains regarding the co-existence of RDF content (the semantic web) and plain web content (html, xml, png) for the same resource. Linked data in the header of an HTML resource serves a different purpose and is used in a different context than linked data that is part of a graph of linked data generally. SELFIE is not going to be able to solve the riddle of unifying JSON-LD for the semantic web and JSON-LD for search engine and other crawler consumption.
There is a lot more work to do to unify these two (RDF content and HTML content with embedded linked data for the same URL). It's my opinion that acting as if we will be able to make recommendations in this space is naive. We can offer findings and make recommendations about some specific technologies in the context of an incremental progression.
The aim of ELFIE, as stated in our charter was:
and the am of SELFIE:
Given this, I want to propose that we focus further discussion in SELFIE as follows.
How additional media-types (RDF-* in particular) play with the core OGC-API media types should be out of scope and we should discuss the difficulties we experienced with this area in the engineering report.
I'm running the risk of making this too long to matter, but one last thing to illustrate this proposal.
In reality, if an implementation were to have an API that could provide an html page or geoJSON data for a given URL, the html would have a
<script>
tag containing JSON-LD inspired by the SELFIE-index context. and the geoJSON would be just that.Any additional capabilities to present the linked data (a JSON-LD or RDF mimetype) or some other linked data for the url is essentially out of scope for SELFIE. While there may be some or even complete overlap between the linked data in the
<script>
tag, there is not necessarily a 1:1 correspondence between the two. To be very clear -- out of scope means we are not going to say we've found answers worth recommending as common practices in this space -- there are many ongoing projects that are working on those answers and we should all take part in them as part of the long term arc of development here.