Open Data and Linked Open Data Session @ SemanticDays
08:30 - 11:30, June 9, 2011
Organized by Dumitru Roman, SINTEF, Norway
Target audience: Researchers, engineers, and students
The Datalift Project: Ontologies, Datasets, Tools and Methodologies to Publish and Interlink Datasets by François Scharffe (University of Montpellier, France)
- Abstract: Publishing linked data can be as simple as installing a tool on a server in a few clicks. However, publishing quality linked data is another story. Ontologies need to be carefully selected and designed, reference datasets need to be identified and the corresponding resources linked to the new dataset. These tasks require both a considerable amount of effort, as well as the availability of reference ontologies and datasets to publish and link the data. The Datalift project proposes to tackle these issues on the example of publishing french public data. Datalift aims at developing a suite of tools easing the publication process, and to provide reference ontologies and datasets that can serve as authoritative datasets to which others can link. We will present the current progress toward reaching this goal, in particular we will focus on Linked Open Vocabularies, a tool for analysing vocabulary usage on the Web of data, and on techniques allowing to automate the data linking process. We will also show our progress towards building a French linked-data ecosystem.
An Overview of Norwegian Linked Open Data by David Norheim (Computas AS, Norway) and Dumitru Roman (SINTEF, Norway)
- Abstract: With Norway being one of the few countries outside of the English speaking world with a clear governmental strategy and commitment to open data, combined with one of the highest Internet penetration and mobile access in Europe, it offers interesting opportunities for becoming a great testbed for consuming Linked Open Data (LOD). With this presentation we aim at presenting potential applications consuming Norwegian LOD and showing practical benefits of aggregating open data in highly sensitive domains for governments and the general public such as regional development and environmental friendly behaviour. These applications will not only aim at demonstrating the benefits of the current Norwegian LOD, but will also make contributions to the improvement and extension of the existing data sets.
How to maintain the primary nature of public sector data: A first foray by Audun Stolpe (University of Oslo, Norway)
- Abstract: Although the World Wide Web and the new media show a bias towards getting the information to flow, the means of assessing and ensuring information integrity is increasingly recognized as a real and pressing need. Public sector information is a case in point: Whilst the political pressure for reusable public sector information is building momentum, governments as authoritative sources of information on the Web must at the same time face the challenges related to maintaining the primary nature of its data. In this talk we shall be concerned with one aspect of this problem, namely the question of what it means, if anything, to reuse data in a non-distortive manner, where distortiveness is measured in relation to the original source. More specifically, we place this problem in the context of RDF and Linked Data, where the principal means of data repurposing is the class of SPARQL queries known as construct-queries.
From Linked Data to Linked Rules by Harold Boley (Institute for Information Technology, National Research Council; Faculty of Computer Science, University of New Brunswick, Canada)
- Abstract: Linked Data (including Linked Open Data and Linked Enterprise Data) have become a new focal point of Semantic Web R&D with potentially high impact on IT at large. Data typically come in the form of relational or graph databases (e.g., triple stores). Relational data can be viewed as ground (variable-free) facts of deductive databases, and facts are rules with an empty premise. Similarly, graph data can be viewed as ground frames of F-logic, RIF, PSOA RuleML, or N3, which can also be defined by rules. Therefore, a natural generalization of Linked Data is Linked Rules. Via bottom-up execution, Linked Rules permit the 'bulk' derivation of Linked Data sets, contexts, modules, or named graphs. Via top-down execution (querying), they permit 'selective' derivations on demand. The linking mechanism of Linked Data and Rules is IRIs (Internationalized Resource Identifiers), which can be used as OIDs (Object Identifiers) of frames, and also occur as slot fillers, slot keys, type identifiers, and document identifiers for modules. Linking between entire data and rule modules is the essence of Linked Data and Linked Rules.