Crowdsourcing Semantic Data Management: Challenges and Opportunities
The paper was published in Proceedings of the International Conference on Web Intelligence, Mining and Semantics (WIMS '12), held in Craiova, Romania on June 13-15, 2012. Elena Simperl was invited as keynote speaker in this conference.
Linked Data refers to a set of guidelines and best practices for publishing and accessing structured data on the Web. It builds upon established Web technologies, in particular HTTP and URIs, extended with Semantic Web representation formats and protocols such as RDF, RDFS, OWL and SPARQL, by which data from different sources can be shared, interconnected and used beyond the application scenarios for which it was originally created. RDF is a central building block of the Linked Data technology stack. It is a graph-based data model based on the idea of making statements about (information and non-information) resources on the Web in terms of triples of the form subject predicate object. The object of any RDF triple may be used in the subject position in other triples, leading to a directed, labeled graph typically referred to as an 'RDF graph'. Both nodes and edges in such graphs are identified via URIs; nodes represent Web resources, while edges stand for attributes of such resources or properties connecting them. Schema information can be expressed using languages such as RDFS and OWL, by which resources can be typed as classes described in terms of domain-specific attributes, properties and constraints. RDF graphs can be natively queried using the query language SPARQL. A SPARQL query is composed of graph patterns and can be stored as RDF triples together with any RDF domain model using SPIN to facilitate the definition of constraints and inference rules in ontologies.