This work describes an infrastructure for the selective RDF data replication to mobile devices while considering current and future information needs of mobile users and the different contexts they are operating in. It presents a novel approach in synthesizing context-aware computing concepts with semantic technologies and distributed transaction management concepts for intelligently assisting mobile users while enhancing mobile information seeking behavior and increasing the precision of mobile information retrieval processes. Despite the huge potential of a proactive, context-dependent replication of RDF data, such data can not be efficiently processed on mobile devices due to (i) technical limitations and network-related constraints, (ii) missing processing and management capabilities of ontology-based description frameworks, (iii) the inability of traditional data replication strategies to adapt to changing user information needs and to consider technical, environmental, and infrastructural restrictions of mobile operating systems, and (iv) the dynamic and emergent nature of context, which requires flexible and extensible description frameworks that allow for elaborating on the semantics of contextual constellations as well as on the relationships that exist between them. As a consequence, existing approaches suffer from the deployment of proprietary data formats, server-dependent application infrastructures, and the inability to share and exchange contextual information across system borders. Moreover, results of recently conducted studies reveal that mobile users find their information needs inadequately addressed, where a large share can be attributed as context or context-relevant. Although progress has been made in applying semantic technologies, concepts, and languages to the domain of context-aware computing, a synthesis of those fields for the proactive provision of RDF data replicas on mobile devices remains an open research issue. This work discusses possible fields where context-aware computing can be enhanced using technologies, languages, and concepts from the Semantic Web and contains a comparative study about the performance of current mobile RDF frameworks in replication-specific tasks. The main contribution of this thesis is a formal description of an abstract model that allows for an efficient acquisition, representation, management, and processing of contextual information while taking into account the peculiarities and operating environments of mobile information systems. It is complemented by a formal specification of a concurrently operating transaction-based processing model that considers completeness and consistency requirements on data and process level. We demonstrate the practicability of the presented approach trough a prototypical implementation of context and data providers that satisfy typical information needs of a mobile knowledge worker. As a consequence, dependencies to external systems are reduced and users are equipped with relevant information that adheres to their information needs anywhere and at any time, independent of any network-related constraints. Since context-relevant data are processed directly on a mobile device, security and privacy issues are preserved.

Top