The Semantic Web enables people and computers to interact and exchange information. Based on Semantic Web technologies, different machine learning applications have been designed. Particularly to emphasize is the possibility to create complex metadata descriptions for any problem domain, based on pre-defined ontologies. In this paper we evaluate the use of a semantic similarity measure based on pre-defined ontologies as an input for a classification analysis. A link prediction between actors of a social network is performed, which could serve as a recommendation system. We measure the prediction performance based on an ontology-based metadata modeling as well as a feature vector modeling. The findings demonstrate that the prediction accuracy based on ontology-based metadata is comparable to traditional approaches and shows that data mining using ontology-based metadata can be considered as a very promising approach.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

Centruflow is an application that can be used to visualise structured data. It does this by drawing graphs, allowing for users to explore information relationships that may not be visible or easily understood otherwise. This helps users to gain a better understanding of their organisation and to communicate more effectively. In earlier versions of Centruflow, it was difficult to develop new functionality as it was built using a relatively unsupported and proprietary visualisation toolkit. In addition, there were major issues surrounding information currency and trust. Something had to be done, and this was a sub-project of this thesis. The main purpose of this thesis however was to research and develop a set of mathematical algorithms to infer implicit relationships in Centruflow data sources. Once these implicit relationships were found, we could make them explicit by showing them within Centruflow. To enable this, relationships were to be calculated based on providing users with the ability to 'tag' resources with metadata. We believed that by using this tagging metadata, Centruflow could offer users far more insight into their own data. Implementing this was not a straight-forward task, as it required a considerable amount of research and development to be undertaken to understand and appreciate technologies that could help us in our goal. Our focus was primarily on technologies and approaches common in the semantic web and 'Web 2.0' areas. By pursuing semantic web technologies, we ensured that Centruflow would be considerably more standards-compliant than it was previously. At the conclusion of our development period, Centruflow had been rather substantially 'retrofitted', with all proprietary technologies replaced with equivalent semantic web technologies. The result of this is that Centruflow is now positioned on the forefront of the semantic web wave, allowing for far more comprehensive and rapid visualisation of a far larger set of readily-available data than what was possible previously. Having implemented all necessary functionality, we validated our approach and were pleased to find that our improvements led to a considerably more intelligent and useful Centruflow application than was previously available. This functionality is now available as part of 'Centruflow 3.0', which will be publicly released in March 2008. Finally, we conclude this thesis with a discussion on the future work that should be undertaken to improve on the current release.

The Semantic Web enables people and computers to interact and exchange information. Based on Semantic Web technologies, different machine learning applications have been designed. Particularly to emphasize is the possibility to create complex metadata descriptions for any problem domain, based on pre-defined ontologies. In this paper we evaluate the use of a semantic similarity measure based on pre-defined ontologies as an input for a classification analysis. A link prediction between actors of a social network is performed, which could serve as a recommendation system. We measure the prediction performance based on an ontology-based metadata modeling as well as a feature vector modeling. The findings demonstrate that the prediction accuracy based on ontology-based metadata is comparable to traditional approaches and shows that data mining using ontology-based metadata can be considered as a very promising approach.

Depuis quelques années, des évolutions importantes ont lieu en matière d’infrastructures technologiques. En particulier, la démocratisation des dispositifs mobiles (comme les PCs, Smartphones, Tablettes, etc.) a rendu l’information accessible par le grand public partout et à tout moment, ce qui est l’origine du concept d’informatique ubiquitaire. L’approche classique des systèmes de l’informatique ubiquitaire, qui répondent aux besoins des utilisateurs indépendants les uns des autres, a été bouleversée par l’introduction de la dimension sociale. Ce rapprochement est à l’origine d’une discipline naissante « le pervasive social computing » ou l’informatique socio-pervasive. Les applications socio-pervasives connaissent une véritable expansion. Ces dernières intègrent de plus en plus la notion de communauté. Le succès des applications communautaires se justifie par le but poursuivi par ces dernières qui est de répondre aux besoins des communautés et d’offrir un ‘chez soi’ virtuel, spécifique à la communauté, dans lequel elle va construire sa propre identité et réaliser ses objectifs. Par ailleurs, la notion de communauté représente une source d’informations contextuelles sociales. Elle est, aujourd’hui, au cœur des problématiques de personnalisation et d’adaptation des applications informatiques. Dans le cadre de cette thèse, nous étudions sous différents aspects les applications informatiques centrées communautés existantes et soulignons un certain nombre de carences au niveau même de la notion de communauté, des modèles de communautés, ou encore des architectures dédiées à ces applications communautaires, etc. Pour remédier à ces défauts, nous proposons trois principales contributions : Un nouveau type de communauté adapté aux exigences des environnements pervasifs qui vient rompre avec les traditionnelles communautés pérennes thématiques : des communautés éphémères, géolocalisées et spontanées (sans contrainte thématique). 
 Un modèle de communauté basé sur les standards du web sémantique pour répondre aux problèmes liés à l’hétérogénéité de conception des communautés. Une architecture dynamiquement reconfigurable pour promouvoir les communautés spontanées en aidant les utilisateurs nomades à intégrer des communautés environnantes et à découvrir les services dédiés. 
Nous montrons la faisabilité de nos propositions pour la conception et le développement d’applications communautaires spontanées grâce au prototype Taldea. Enfin, nous testons les approches proposées de découverte de communauté et de services à travers plusieurs scénarios caractérisés par la mobilité et l’ubiquité. / Advances in technology, in particular the democratization of mobile devices (PCs, smartphones and tablets), has made information accessible to anyone at any time and from anywhere while facilitating the capture of physical contextual data, thereby justifying the growing interest for pervasive computing. The classical approach of pervasive computing has been affected by the introduction of the social dimension. Ubiquitous systems do not meet the needs of users independently from each other but do take into account their social context. Fostering the social dimension has given rise to a fast growing research field called Pervasive Social Computing. Applications in this area are increasingly concerned by communities. The contextual information associated with a community can be harnessed for personalization, adaptability and dynamic deployment of services, which are important factors for Pervasive Computing. A community is considered in our approach as a set of distinct social entities that should be supported with services as a single user is. In this thesis, we look into different aspects of existing centered communities applications and we identify several weaknesses and shortcomings in the notion of community, the community models, and the architecture of communities’ applications. To overcome these shortcomings, we propose three main contributions: A new type of communities that fits better with the requirements of pervasive environments: short- lived, geolocated and spontaneous (without thematic constraint) community. Intuitively, it is the type of community that best matches with circumstantial, accidental, incidental or fortuitous situations. This kind of community has to meet specific needs, which are not taken into account by perennial thematic communities. 
 A model for communities based on semantic web standards to overcome the problem of heterogeneity across definitions and models. The ontological representation allows us to organize and represent social data, to make information searches easier for users and to infer new knowledge. 
 A dynamically reconfigurable architecture for fostering spontaneous communities in order to facilitate the user access to communities, information exchange between community members and service discovery. 
The proposed architecture for community and service discovery have been validated through a prototype called Taldea and have been tested through several scenarios characterized by mobility and ubiquity.