In a world of big, unstructured and volunteered data, one of the major challenges is to integrate or link the emergent datasets together, so that they can be published and put to use. As an example, consider the problem of recognizing and gathering together information generated on social media relating to a forest fire, a flood or even a football game. The semantic web is heavily dependent on vocabularies and ontologies by which to describe data, but often these do not exist; and when they do exist they are often inadequate, out of date or private.
VoCamps are a grass-roots movement that has arisen to tackle this problem by creating open, lightweight, reusable semantic patterns that can parse, structure and harmonise unstructured data. These ‘Volunteered Ontologies’ are created by expert groups in an unconference setting and are then freely published. The compromise approach that characterizes VoCamps is to try to create small, reusable ontological patterns that have high utility, but are easy to apply and to understand. This way we can 'capture' more linked data resources without massive ontological engineering efforts that may not be sustainable and often lead to unplanned restrictions in semantic heterogeneity.
This talk provides a brief introduction to some of the data challenges of the Semantic Web, discusses three approaches to develop the missing semantics and outlines why the Volunteered Ontology approach is enjoying some success. Examples are provided of work undertaken at recent VoCamps in the spatial domain to show how the pattern creation process works, and how the resulting patterns can harness and put to use large, unstructured data collections.