The Semantic Tool For Screen Arts Research (STARS) is a project of the University of Bristol’s Department of Drama: Theatre, Film, Television and The Institute for Learning and Research Technology and with (the very cool looking) Watershed’s dShed.
STARS is a web app that brings together a number of ‘semantic web’ tools and interesting audio visual datasets in order to benchtest the potential for developing an open ended space for annotating audio visual material. That said the assets STARs is capable of working with extend well beyond the audio visual.
While STARS seems particularly well suited and uniquely capable for the collaborative annotation of rich media – the possibilites it presents extend well beyond this single facility. The real value of STARS lies in the model it presents for collaboratively mapping and actively developing a dynamic space of richly connected and widely varied assets – rich media, institutions, people, concepts, projects, events, text, taxonomies and folksonomies (annotations).
STARS allows a user to search any of the prescribed datasets via keyword or specified filter. It returns results identified by a neat icon key that identifies them by those varied asset types. The search provides a brief description, an option to reveal an detail description and semantic components, and an option to open a ‘mapping’ of the item.
Opening the map reveals a visualisation of the items semantic connections in a number of varied diagrams (linear, cubic, distrubuted). In each case the map provides an interesting description of the relation between associated asset types. The most obvious example might be a map centered on an institution that has a number of people attached – has links to other institutions through projects – etc. These maps all open onto semantic descriptions which can be further annotated. I imagine these maps getting much more interesting when video annotations start mapping memes or technical qualities throughout a dataset. The great thing about STARs is that it has kept the annotations, assets and so on on the same level as assets of the order of institution and people. A completely flat ontology like this is incredibly powerful – infinitely generative – because it refrains from prescribing a hierarchy or limit the way things, bodies, concepts, assemblages potentially interact – In fact these very interactions become assets in the database – no longer ‘meta’ – they’ve become differential.
For instance with a system like this it becomes plausible that you might realise oblique connections between otherwise disparate researchers via the way their varied taxonomies are applied in a set of like media annotations.