Following my intuition keeps leading me to fun coincidences.
I've been following the work of Dorian Taylor, learning more about how he uses linked data. App::IBIS is interesting, but also how he's using attributes on links on his website.
Here's an example:
<a href="https://doriantaylor.com/policy/uris-resources-and-representations" rev="dct:references xhv:up" typeof="bibo:Report"> <span property="dct:title"> URIs, Resources and Representations </span> </a>
I've implemented a markdown syntax so that I could assign arbitrary attributes on links. If I wanted to do something like this, that's the way I'd have to go. If I only used
rev="dct:references" it's simpler but still ugly, and it doesn't provide any context for someone using a web browser.
Today on a Federated Wiki Zoom, there was a demo showing how Eric Dobbs labels the relations between pages using Graphviz DOT language.
He uses a syntax that matches the words that start a line that link to other pages. Examples are "Includes," "Consists of," or "Enabled by." I could see this used in conjunction with the work Dorian is doing. I could prefix a line "References," and my blog engine would know to add
rev="dct:references" to all the links on that line.
Pulling from his Content Inventory I could prefix links with "Mentions," "Introduces," or "Evokes."
I'm excited about this idea. I think the next step is to decide my use case. What types of links do I want? Do I want to have matched inverses? So any page linked with "Evokes" would automatically have a backlink that is "Evoked by."
The link prefix context is way more relevant to what I'm doing because I want it to be useful to a human, not necessarily a scraper.
Published December 23, 2020