Linkage in haze: challenges and take-home messages of crowd-sourcing vagueness in musical data
Alessandro Adamou, The Open University
With the transition of the Web of today from an information repository to a suite of services, the demand for machine-readable data to support the latter is now greater than ever. The social and, more generally, community element is proving to be a valuable medium to convey such a bulk of knowledge. Linked Data is a leading body of standards for publishing and using open knowledge bases on the Web, however, it very much relies upon the notion of identity. Every object of the world being described should be uniquely identified in order to be effectively manipulated. Music is a specially provocative domain of interest for such Web knowledge bases, being a topic where most people feel confident they can contribute to, yet with varying degrees of factual knowledge, personal inclination or scholarly rigour. Curating a dataset that covers an aspect new to this landscape, as is the evidence of listening experiences, means dealing with partial, inexplicit or underspecified information. A likely implication is that several elements of a listening experience, such as the listeners, the time in history or the music being heard, can be described to an extent but not identified, thus in stark contrast with a founding principle of Linked Data. This talk will illustrate the nature of the main elements of fuzzy knowledge that emerged from the contributions to the Listening Experience Database, elaborate on the countermeasures adopted and lessons learnt from the life-cycle of LED data, and assess the state of maturity of Linked Data technologies for accommodating such use-cases.