Les travaux présentés dans cette thèse proposent différentes méthodes, réflexions et réalisations associant Web 2.0 et Web Sémantique. Après avoir introduit ces deux notions, nous présentons les limites actuelles de certains outils, comme les blogs ou les wikis, et des pratiques de tagging dans un contexte d’Entreprise 2.0. Nous proposons ensuite la méthode SemSLATES et la vision globale d’une architecture de médiation reposant sur les standards du Web Sémantique (langages, modèles, outils et protocoles) pour pallier à ces limites. Nous détaillons par la suite différentes ontologies (au sens informatique) développées pour mener à bien cette vision : d’une part, en contribuant activement au projet SIOC - Semantically-Interlinked Online Communities -, des modèles destinés aux méta-données socio-structurelles, d’autre part des modèles, étendant des ontologies publiques, destinés aux données métier. De plus, la définition de l’ontologie MOAT - Meaning Of A Tag - nous permet de coupler la souplesse du tagging et la puissance de l'indexation à base d’ontologies. Nous revenons ensuite sur différentes implémentations logicielles que nous avons mises en place à EDF R&D pour permettre de manière intuitive la production et l'utilisation d'annotations sémantiques afin d’enrichir les outils initiaux : wikis sémantiques, interfaces avancées de visualisation (navigation à facettes, mash-up sémantique, etc.) et moteur de recherche sémantique. Plusieurs contributions ont été publiées sous forme d'ontologies publiques ou de logiciels libres, contribuant de manière plus large à cette convergence entre Web 2.0 et Web Sémantique non seulement en entreprise mais sur le Web dans son ensemble. / The work described in this thesis provides different methods, thoughts and implementations combining Web 2.0 and the Semantic Web. After introducing those terms, we present the current shortcomings of tools such as blogs and wikis as well as tagging practices in an Enterprise 2.0 context. We define the SemSLATES methodology and the global vision of a middleware architecture based on Semantic Web technologies (languages, models, tools and protocols) to solve these issues. Then, we detail the various ontologies (as in computer science) that we build to achieve this goal: on the one hand models dedicated to socio-structural meta-data, by actively contributing to SIOC - Semantically-Interlinked Online Communities -, and on the other hands models extending public ontologies for domain data. Moreover, the MOAT ontology - Meaning Of A Tag – allows us to combine the flexibility of tagging and the power of ontology-based indexing. We then describe several software implementations, at EDF R&D, dedicated to easily produce and use semantic annotations to enrich original tools: semantic wikis, advanced visualization interfaces (faceted browsing, semantic mash-ups, etc.) and a semantic search engine. Several contributions have been published as public ontologies or open-source software, contributing more generally to this convergence between Web 2.0 and the Semantic Web, not only in enterprise but on the Web as a whole.

No description available.

Sub-micron patterning is fundamental to the fabrication of numerous devices Traditional commercial manufacturing methods either lack the resolution needed to attain the appropriate size or are prohibitively expensive due to low throughput or the necessity of expensive equipment. Imprint lithography is a rapid, inexpensive alternative to making sub-micron features that can be tailored to work with a variety of materials. Imprint lithography, while traditionally used with pure polymers has been tailored to be used with nanoparticle-polymer hybrid films. This work has achieved high-fidelity pattern transfer onto polymer-nanoparticle hybrid films with feature sizes as small as 250 nm. The polymer-nanoparticle hybrid was fabricated by creating a liquid suspension of functionalized ZnO nanoparticles and poly(methyl methacrylate) (PMMA) in a solvent. The ZnO particles were functionalized by adding nonanoic acid in order to facilitate the dispersion of the particles in a non-polar solvent. This suspension was spread onto substrate, imprinted with a patterned stamp, allowed to dry, and was demolded. The final result was features ranging from 250 nm to 1 μm in size with good fidelity as determined by the accuracy of the feature replication and the surface roughness of the overall sample. The effect of the ZnO content as well as the method of combining the suspension components on the feature fidelity was studied. In general, it was found that feature fidelity is acceptable up to a dry-film composition of 15 vol% ZnO and that feature sizes above 500 nm were more tolerant of higher solids loading. The same imprint lithography method was also used to pattern a polymer-derived SiOC glass. The SiOC was shown to be have interesting shrinkage properties where the feature-level linear shrinkage was up to 5% more than that of the bulk. The features were shown to be stable during pyrolysis up to 1000°C and stable at operating temperatures up to 1000°C. A constant number Monte Carlo simulation was used to describe the suspension behavior to confirm the empirical results from the physical experiments. The effects of Van der Waals forces, steric stabilization, depletion flocculation, as well as the physical impediment of entangled polymer chains were considered. A similar agglomeration behavior was shown in the simulations compared to the physical experiments. This thesis shows that polymer-nanoparticle hybrid films are a compatible material for imprint lithography using appropriate suspension parameters. This is very important for a variety of applications and devices. Using imprint lithography to make these devices makes them cheaper and more accessible to the commercial market and can make a large number of theoretical devices a reality. / Ph. D. / Sub-micron patterning is an integral part of making many modern technologies such as memory storage devices or integrated circuits. As this technology becomes smaller and smaller, the limiting factor for making these devices has become the ability to manufacture effectively at the appropriate scale. Traditional commercial manufacturing methods lack the resolution needed to attain small enough features. Manufacturing methods that can make small enough features are often either extremely expensive or offer incomplete control of the feature morphology. Imprint lithography is a high-throughput, inexpensive alternative to making sub-micron features that can be tailored to work with a variety of materials. Imprint lithography is simple process in which a patterned stamp is pressed into a softened film of material in order to transfer the pattern of the stamp onto that material. Traditionally, imprint lithography works best with polymers and researchers have struggled to pattern nanoparticle-based materials. This work has achieved high-fidelity pattern transfer onto polymer-nanoparticle hybrid films with feature sizes on the same order as the polymer films found reported in literature. The polymer-nanoparticle hybrid was realized by creating a liquid suspension of functionalized ZnO nanoparticles and poly(methyl methacrylate) (PMMA) in a solvent. The ZnO particles were functionalized by adding nonanoic acid, allowing the normally polar particles to disperse in the non-polar solvent needed to dissolve the PMMA. This suspension was spread onto a glass substrate, imprinted with a patterned stamp, allowed to dry, and was demolded. The final result was the successful transfer of features ranging from 250 nm to 1 μm in size with good fidelity. The effect of the ZnO content as well as the method of combining the suspension components on the feature fidelity was studied. To help prove the broad applicability of this imprint method, it was adapted for use with polymer-derived ceramics. Additionally, a computer simulation was developed to help understand the behavior of the nanoparticle-polymer suspension during the imprint process.