The past few weeks of DITA have been interestingly technical. We are so surrounded by computerised technology that it is easy to take for granted the code that underpins it. When I think back on my childhood, I recall primary school computer classes, learning BASIC on an Apple II. Although at that age, we’d seen the move from cassette tapes to floppy discs, I had no reason to place them in a lineage of punch-carded commands. And although it was exciting to learn how to paint the screen a given colour, it wasn’t too long after that the graphical user interface and its painting applications made that ‘skill’ obsolete. So it has been enlightening and useful to put that heritage, and our current technological moment, into some historical context.
That said, if not directly programming, my career has tended to involve technology; it’s hard to avoid in any profession, but data and systems are fundamental to financial management. Even as I was training to be an accountant, I was using Lotus 1–2–3, linking worksheets using formulas to aggregate the granular into a more streamlined summary. I mightn’t have known the terminology, but of course that meant creating miniature databases, and over time, I gained exposure to more complex relational ones. Indeed, I work with one now on a daily basis, as both a user and architect. When planning any development effort, the aim is to normalise the data as much as possible, establishing a relationship framework that allows dimensions of entries to organised for searching and reporting.
Given that background, introduction to the Resource Description Framework (RDF) as a data model has been revelatory. I am accustomed to interrogating predictably arranged tables. And yet, as I grapple with the notion of linked data and the semantic web, I imagine triples’ amorphous variation and apparently infinite options. I can understand ontologies help to tame, or at least define, those relationships in a managed way, but a major challenge is working out how.
It’s not clear to me that problem has been satisfactorily solved. Looking at schema.org’s definition of the properties for which instances of Thing may appear provides an amusing insight into the development of a supposedly universal approach: 17% of them relate to tabletop or video games. Not to cast aspersions on the important work of the community that develops the structure, but the joke sort of writes itself.
But if the generic won’t do, there’s the interesting question of what will. Lynnsey Weissenberger’s presentation on the Linked Irish Traditional Music project at DocPerform2 elaborated on the challenges of using a conventional music ontology to represent the complexity she found in the arena of traditional music. She highlighted the importance of foregrounding humans in the documentation of the material, but looking at the Person properties of schema.org’s MusicRecording, the shortcomings for her project become immediately apparent. Even if there is a recording of the material, the notion of, for example, either a “creator/author” or a “Person that is legally accountable for the CreativeWork” is in many respects contrary to the way music is understood by its makers or audience.
That is but one illustration, but it seems a case in point about the challenge we face when it comes to moving to a world of linked data. Being a walled garden can certainly be seen as a shortcoming of a standalone relational database. Nevertheless, given confines, one can at least attempt to keep that garden tidy. Hyperlinks obviously introduce some openings in the wall: windows, if you will, or doors. But the logical extension of a linked model is a garden without any walls and only vague boundaries. That vision could be one of a utopian plain, but without some consistency in the mapping, it could also be a Wild West. Weissenberger acts as one cartographer, exploring a path toward a methodology that can properly describe her discipline.
Still, the ultimate struggle is a macro-approach that situates it, and its attendant idiosyncrasies, within a broader framework of many other disciplines. Those are apt to have Persons, Places and Things in common — there are only so many types of subjects and objects, after all, and the Dublin Core elements provide something of a lowest common denominator for them. (The Things, at least.) But their relationships are the tricky thing. A macro-schema may be possible, but it is difficult to imagine. That said, if one does emerge, its taxonomy seems unlikely to give video games quite the same degree of prominence.
Image: screenshot of Google Images search on ‘linked data‘