raimond.me.uk
Yves Raimond's home

A set of OWL-DL ontologies designed to cover a large range of things happening during a music production process. This ontology is split in several ones.

The Music Ontology homepage is now HERE. We describe here some ontologies that are being used by the Music Ontology and that are hosted on this server.

Time ontology

We needed to write another time ontology, as most of the one available in core ontologies (DOLCE) were not expressive enough to express complex relationships between several temporal objects (like the one between a given performance, the continuous signal recorded, the sampled signal, and the archive representing it). Thus, this ontology defines time-lines, acting as a backbone for time intervals (represented as in Allen's theory) or time points, and time-line maps allowing to express the relationship between two of them.

With this knowledge, we are able to infer many many interesting things, like getting back from a particular audio file to the exact time, on the universal time-line, to which a particular sample corresponds to. Or even being able to adress a particular time interval using a continuous time-line on an archived digital signal.

Event ontology

By defining a new time ontology, we also had to write our own event ontology. This ontology is based on the token reification approach. Basically, the event acts as an arbitrary way of classifying space/time regions. An event can be linked to either agents (active participants) or factors (like a particular instrument), and obviously, products. Moreover, we introduce the transitive property hasSubEvent which allows to split a complex event into less complex one. For example, we can express, for a whole music performance, that a particular person was playing a particular instrument at a particular time, and using the time ontology, we can map this particular sub-event to the corresponding interval on the archived signal.

Amazon's genre taxonomy in RDF

I wrote a small hack in order to grab their (sort-of) genre taxonomy from their XML web-services and convert it into RDF (non-strict subClassOf hierarchy). The result is a bit... hmmm... chaotic, but nevertheless it is funny:-) By the way, there is lots of data available that are annotated against this "taxonomy".