First method: Loose ethnography on the birth of ontologies
We aim to research in our London ethnographic fieldwork the ‘ birth’ of ontologies. Ontologies order metadata, stimulate its production, facilitate its exchange and ease its usage. In a way they can be described as the ‘ genetic code’ of the new semantic web. Ontologies are normally born in projects that involve worldwide collaboration. They are thus not really confined to one particular place. However, London and the South of England seem to us as a particular hub for this (Tim Berners Lee, to take the most obvious example, is at last partly institutionally based at the University of Southampton).
Thus we have decided to research this side of our project – the early days of new ontologies, largely in Britain. The second reason to pursue this side of the project in the UK is that we aim to follow the early days of such ontologies over a longer period of time. 18 months, to be more precise. As we are based in London, we can use here a form of media ethnography that is it looser and more long-term.
However, the process of the early development of such ontologies of course extends the 18 months. We have therefore decided to conduct our long-term ethnographic fieldwork in four settings. Each represents a certain moment of time in the process of the come-to-being of ontologies:
(1) We aim to observe one of the working groups of W3C, the World Wide Web Consortium. W3C promotes the semantic web, and sets up recommendations. We are interested in W3C, because this organisation sets the frame for the development of ontologies.
(2.) Transmission.cc is an international network of citizen journalists, programmers and video makers, who try to set up a metadata standard for independent audiovisual material, thus improving the search and the accessibility of such material. This group aims to create a new ontology, and it is still pretty early days.
(3.) The Music Ontology (MO), developed by Yves Raimond et al., is a new ontology for music-related metadata that aims to include all descriptions related to music. Currently this ontology has come to a relatively stable version, and will be used by the large music metadata sharing website MusicBrainz in the near future. We can therefore observe the (relative) stabilisation of an ontology.
(4.) Omras2 (Online Music Recognition And Searching). This research project develops a new form of metadata management for music files. It will store more complex information (e.g. on the preferences of other listeners) and will thus enable, for example, smart playlists for music, comparable to a desktop-based form of last.fm. In doing so, it will implement the Music Ontology that was developed by Yves Raimond (see above). We will take part in the process of its implementation.
All these projects are rather loose networks of academics, activists, metadata geeks and stakeholders in the industry form. Some of them have never met in person. We will follow all four case studies for 18 months. We will conduct interviews with some of the England-based developers; follow the mailing lists; attend meetings, workshops, and events; and, possibly, and to some extend, also participating in their production process.