The organization of (multimedia) data on current desktop systems is done to a large part by arranging files in hierarchical file systems, but also by specialized applications (e.g., music or photo organizing software) that make use of file-related metadata for this task. These metadata are predominantly stored in embedded file headers, using a magnitude of mainly proprietary formats. Generally, metadata and links play the key roles in advanced data organization concepts. Their limited support in prevalent file system implementations, however, hinders the adoption of such concepts on the desktop: First, non-uniform access interfaces require metadata consuming applications to understand both a file’s format and its metadata scheme; second, separate data/metadata access is not possible, and third, metadata cannot be attached to multiple files or to file folders although the latter are the primary constructs for file organization. As a consequence of this, current desktops suffer, inter alia, from (i) limited data organization possibilities, (ii) limited navigability, (iii) limited data findability, and (iv) metadata fragmentation. Although there were attempts to improve this situation, e.g., by introducing semantic file systems, most of these issues were successfully addressed and solved in the Web and in particular in the Semantic Web and reusing these solutions on the desktop, a central hub of data and metadata manipulation, is clearly desirable. In this thesis a novel, backwards-compatible metadata model that addresses the above-mentioned issues is introduced. This model is based on stable file identifiers and external, file-related, semantic metadata descriptions that are represented using the generic RDF graph model. Descriptions are accessible via a uniform Linked Data interface and can be linked with other descriptions and resources. In particular, this model enables semantic linking between local file system objects and remote resources on the Web or the emerging Web of Data, thereby enabling the integration of these data spaces. As the model crucially relies on the stability of these links, we contribute two algorithms that preserve their integrity in local and in remote environments. This means that links between file system objects, metadata descriptions and remote resources do not break even if their addresses change, e.g., when files are moved or Linked Data resources are re-published using different URIs. Finally, we contribute a prototypical implementation of the proposed metadata model that demonstrates how these building blocks sum up to constitute a metadata layer that may act as a foundation for semantic data organization on the desktop.

Top