On reflection in Linked Data management

Abstract

The Linked Data (LD) initiative promotes the use of international standards for data sharing on the web. The flexible Resource Description Framework (RDF) graph data model is a central LD technology. Much of RDF's flexibility is due to its blurring of the traditional distinction between data and metadata. A particularly powerful form of metadata is query expressions. A query language is called reflective if it can interpret expressions of the language stored as data both as “active” data and as regular “static” data. Fundamental applications of reflection are found in diverse domains such as security, information integration, and data quality. Consequently, a variety of RDF representations of active data have been proposed. However, RDF query languages, while bridging the data-metadata barrier, continue to maintain a divide between active and static data. Hence, applications of RDF active data rely on ad-hoc special-purpose solutions, thereby limiting their broader use and impact. This paper argues that active data must be promoted as first class citizens in RDF querying, if we are to realize the full potential of LD. The aim is to stimulate the study of general frameworks and solutions for reasoning about and applying reflection in the web of data, towards addressing this fundamental research challenge.