celebrating 20 years of vapourware
I'm currently rearranging things. This is actually the history.
Current activity is at : https://hyperdata.it/hkms/
Hyperdata Knowledge Management System (HKMS) is the current working title of a personal project I've been working on for a long time.
Skip down to 'Components' if you wish to avoid the ramblings of an old man.
HKMS's first incarnation was Ideagraph, intended as a Personal Knowledge Management System.
Its purpose was :
- idea brainstorming and development
- idea communication
- knowledge sharing
- project planning
- document organization
- semantic web blogging
- syndicated news reading, processing and republishing
That might sound like an outrageously wide scope, especially for a lone coder. It was, but it was surprising how far I got in a relatively short period of time as an unexceptional coder. The nearest it got to being finished was the desktop Java app in the screenshot above (other screenshots).
The reason this even remotely possible was that all of these things could share many common components. Critically, they could all share the same data model which was compatible with the Web, by design.
A little personal history to give the above context. I'd been interested in AI since seeing 2001 A Space Odyssey as a kid, reading Asimov books. I got into electronics and computing as a teenager. My first computer was a Commodore PET on which I first played with AI : a Tick-Tack-Toe (Noughts & Crosses) player, typed in from a magazine. Years later, at university, I took modules in neural nets and expert systems. Later still, I got online and had a day job at a college that included looking after the network, user support, database admin etc. But it did give me machines to play with
Around 1999 I stumbled upon RDF. This seemed to connect the dots between the interesting AI bits (for me at the time) and the Web. At the time I was clueless at both ends, but the rdf-dev mailing list was very friendly, and I gradually picked bits up.
Sooo...the Resource Description Framework RDF is a set of models and syntaxes for expressing arbitrary data in a form that is natively compatible with the Web. A clue is in the name : Resource in this context is the same as the R in URL. A resource is simply anything that you can identify. A real-world thing, a concept, or more typically on the Web, a document. The neat trick that RDF brings to the table is that you can also identify relationships between things and treat them as resources in their own rights.
Given a model that can identify things and describe relationships between them, you can build networks of information, more commonly known as knowledge graphs. These are versatile enough to be able to express pretty much any can of data you might wish to use on a computer. The practicalities of this can get a bit involved depending on what you're trying to do, but conceptually it's trivial.
Looking back at the list I had for Ideagraph : ideas, projects, blogs, documents... - all things expressible in RDF. Together, standard Java (using Swing for UI) and the nascent Apache Jena framework had all the tools I needed to code something up.
RDF provides a model that can represent pretty much anything. It has various serialization formats so the data can be put in text. But like CSV files, this isn't very user-friendly.
So I wanted to make visualizing and editing RDF a lot more user friendly. In Ideagraph I had three main ways of looking at the RDF:
- text - view source of a particular bit of the RDF, or view/edit a text description that was a property of a node
- tree - a collapsible tree view, similar to those used in file explorers (loops would just keep expanding)
- graph - a node & arc nextwork view
I can't remember how far I'd got with implementation, but I also had more dedicated views for :
- Project - a view structure around a vocab designed for describing projects
- Channels - for RSS feeds
- Vocabs - for RDFS
- Commands - an interpreter (I forget what I used for this, may have been an ad hoc DSL or some standard language interpreter)
- Settings
scope drift
too generic
-
Gradino : RDF-backed blog engine, Java/Scala (retired)
-
Seki : (retired)
-
Scute : a Semantic Web hacking tool, desktop Java (retired)
- NewsMonitor : quasi-intelligent headless feedreader/aggregator
- FooWiki : a browser-based, SPARQL Server-backed Wiki
- Thiki : RDF-backed personal Wiki for Android devices
- Foolicious : a SPARQL-backed bookmark manager (del.icio.us clone)
- SparqlPress
- Trellis : SPARQL-backed hierarchical to-do list
- Schema Editor : browser-based RDF schema authoring tool
- Diced - Getting Things Done with dice
SPARQL Diamonds (github) is an architectural pattern that arose from working on these I'm now trying to use consistently across all the above.
Bookmark manager : the one in eg. Chrome browser is absolutely hopeless