No description available.

Data about the spatial variation of soil attributes is required to address a great number of environmental issues, such as improving water quality, flood mitigation, and determining the effects of the terrestrial carbon cycle. The need for a continuum of soils data is problematic, as it is only possible to observe soil attributes at a limited number of locations, beyond which, prediction is required. There is, however, disparity between the way in which much of the existing information about soil is recorded and the format in which the data is required. There are two primary methods of representing the variation in soil properties, as a set of distinct classes or as a continuum. The former is how the variation in soils has been recorded historically by the soil survey, whereas the latter is how soils data is typically required. One solution to this issue is to use a soil-landscape modelling approach which relates the soil to the wider landscape (including topography, land-use, geology and climatic conditions) using a statistical model. In this study, the soil-landscape modelling approach has been applied to the prediction of soil bulk density (Db). The original contribution to knowledge of the study is demonstrating that producing a continuous surface of Db using a soil-landscape modelling approach is that a viable alternative to the ‘classification’ approach which is most frequently used. The benefit of this method is shown in relation to the prediction of soil carbon stocks, which can be predicted more accurately and with less uncertainty. The second part of this study concerns the inclusion of expert knowledge within the soil-landscape modelling approach. The statistical modelling approaches used to predict Db are data driven, hence it is difficult to interpret the processes which the model represents. In this study, expert knowledge is used to predict Db within a Bayesian network modelling framework, which structures knowledge in terms of probability. This approach creates models which can be more easily interpreted and consequently facilitate knowledge discovery, it also provides a method for expert knowledge to be used as a proxy for empirical data. The contribution to knowledge of this section of the study is twofold, firstly, that Bayesian networks can be used as tools for data-mining to predict a continuous soil attribute such as Db and that in lieu of data, expert knowledge can be used to accurately predict landscape-scale trends in the variation of Db using a Bayesian modelling approach.

Le succès du Web de Données repose largement sur notre capacité à atteindre les données stockées dans des silos invisibles du web. Dans les 15 dernières années, des travaux ont entrepris d’exposer divers types de données structurées au format RDF. Dans le même temps, le marché des bases de données (BdD) est devenu très hétérogène avec le succès massif des BdD NoSQL. Celles-ci sont potentiellement d’importants fournisseurs de données liées. Aussi, l’objectif de cette thèse est de permettre l’intégration en RDF de sources de données hétérogènes, et notamment d'alimenter le Web de Données avec les données issues des BdD NoSQL. Nous proposons un langage générique, xR2RML, pour décrire le mapping de sources hétérogènes vers une représentation RDF arbitraire. Ce langage étend des travaux précédents sur la traduction de sources relationnelles, CSV/TSV et XML en RDF. Sur cette base, nous proposons soit de matérialiser les données RDF, soit d'évaluer dynamiquement des requêtes SPARQL sur la base native. Dans ce dernier cas, nous proposons une approche en deux étapes : (i) traduction d’une requête SPARQL en une requête pivot, abstraite, en se basant sur le mapping xR2RML ; (ii) traduction de la requête abstraite en une requête concrète, prenant en compte les spécificités du langage de requête de la BdD cible. Un souci particulier est apporté à l'optimisation des requêtes, aux niveaux abstrait et concret. Nous démontrons l’applicabilité de notre approche via un prototype pour la populaire base MongoDB. Nous avons validé la méthode dans un cas d’utilisation réel issu du domaine des humanités numériques. / To a great extent, the success of the Web of Data depends on the ability to reach out legacy data locked in silos inaccessible from the web. In the last 15 years, various works have tackled the problem of exposing various structured data in the Resource Description Format (RDF). Meanwhile, the overwhelming success of NoSQL databases has made the database landscape more diverse than ever. NoSQL databases are strong potential contributors of valuable linked open data. Hence, the object of this thesis is to enable RDF-based data integration over heterogeneous data sources and, in particular, to harness NoSQL databases to populate the Web of Data. We propose a generic mapping language, xR2RML, to describe the mapping of heterogeneous data sources into an arbitrary RDF representation. xR2RML relies on and extends previous works on the translation of RDBs, CSV/TSV and XML into RDF. With such an xR2RML mapping, we propose either to materialize RDF data or to dynamically evaluate SPARQL queries on the native database. In the latter, we follow a two-step approach. The first step performs the translation of a SPARQL query into a pivot abstract query based on the xR2RML mapping of the target database to RDF. In the second step, the abstract query is translated into a concrete query, taking into account the specificities of the database query language. Great care is taken of the query optimization opportunities, both at the abstract and the concrete levels. To demonstrate the effectiveness of our approach, we have developed a prototype implementation for MongoDB, the popular NoSQL document store. We have validated the method using a real-life use case in Digital Humanities.