Structuring Factoids in the Dictionary of Sydney

Canberra.
OK, I skipped the first afternoon session at Digital Humanities Australasia 2012 for a quick excursion to Parliament House – and thanks to the vagaries of Canberra taxi services, missed half of the first paper in the next session as well. So, we pick up again with Stewart Wallace from the Dictionary of Sydney project. The dictionary contains some 22,000 entities called ‘factoids’, linked together in various ways.

As a project, it is about urban history; it began with the work of historians, and attempts to reflect their journeys, and to connect their knowledge. The underlying architecture comprises a single digital repository, containing various entities and their interrelations, and presenting these in XML format to be used in various forms of presentation and visualisation.

Underlying this is a distinction between entities, resources, and subjects – so the system is not simply resource-centric, but term-centric. For effective use, knowledge organisation structures must also be kept shallow, so that browsing the resource remains straightforward; additionally, the definition of terms is also time-dependent and changeable.

Overall, then, factoids provide information about ‘things’; this is time-dependent; these factoids are connected through relationships; and these are annotated with ‘roles’ (e.g. type, milestone, name, occupation, relationship, or property – this applies both to people and objects, of course). As new entities are added to the dictionary, this also ‘thickens’ the overall network of factoids it contains, then.

This fundamentally changes the underlying network structure, from a graph centred around relationships between entities to one centred around factoids which are associated with one another in various forms. Factoids have also turned out to be a very useful checking and triangulation device for information contained in the dictionary.