inContext-Sensing is a web application that shows how users with no expertise can benefit of a Linked Data representation to make sense of raw sensor data. In fact average users are becoming the main consumers of sensor data, but sensors conceptualization do not consider their point of view.
The novelty consists in the dynamicity of the linked sensor data creation, since in other state of the art applications the data sets to be linked are usually predefined.
inContext-Sensing is a web application characterized by the following functionalities: (1) automated annotation as RDF of the raw data provided by Pachube about sensor feeds; (2) customized search performed over the LOD cloud for external data that can be linked with a particular observed feature (e.g. Noise, Air quality) selected among those belonging to a Pachube feed. This search is customized since the user can graphically select (a) one or more domains to which the external resources to be found should belong. Specifically the available domains are: government, geography, cross-domain, life science, publication, media, user-generated content, geography; (b) the context which the external resource to be found should share with the selected observed feature. A context is meant to be represented by Space, Time and Thing (of interest), so the user can choose among one of such three options; (c) the confidence level threshold above which the external resources to be found should be ranked (according to the relatedness with the selected observed feature). The search is performed through a query forward to Sindice. The results are shown divided per domain to which they belong while the dataset from which they come from is also displayed.

A DEMO paper has been published as a result, at the International Semantic Web Conference 2011 (ISWC2011): inContext Sensing: LOD augmented sensor data