Most of the Alph software sprang into existence last fall (2016), with our own metadata formats and interfaces, and with our own linking models still on the drawing board.
When the W3C delivered the Web Annotation Data Model recommendation in February of this year – a spec that describes something very much like a xanalogical linking model for Web resources – we took a developmental pause to read into its implications for our own work.
We like Linked Data and JSON-LD. And, we like Web Annotations.
So, alph.py and alph.js are undergoing a transition to work within the Linked Data ecosystem.
The ?describe interface will remain available, but the return format will now be JSON-LD. Our vocabulary is still being ironed-out, but as it develops it will be made available at http://alph.io/terms.jsonld. Media metadata responses are presently modeled on the Schema.org MediaObject types.
The response document for ?link queries will also be JSON-LD, though its structure is still in flux.
Furthermore, resources now send the "Link:" header indicating that they are Linked Data Platform Resources, and will respond with metadata in JSON-LD, RDF, Turtle, or N-Triples if the according "Accept:" header is sent.
With regards to linking, it is our intention to support import/export of Web Annotations. We're very close to being able to do that, in a limited way. We're currently still debating the merits of continuing to work on our own link document format/model. We may opt instead to extend the Web Annotation vocabulary slightly.
Some documentation is being written for our terse fragment selectors, and the way in which we've implemented the element(), point(), and range() XPointer schemes.
Documentation generally is, we fully acknowledge, nonexistent. Frankly, as "we" are one man (howdy!), and as Alph is at this time a personal project, it will remain a low priority until we've got something working well enough to try and attract interest from the wider world.