With the popularity of the World Wide Web, a large amount of data is generated and made available through the Internet everyday. To integrate and query this huge amount of heterogeneous data, the vision of Semantic Web has been recognized as a possible solution. One key technology for the Semantic Web is ontologies. Many ontologies have been developed in recent years. Meanwhile, due to the demand of applications using multiple ontologies,  mappings between entities of these ontologies are generated as well, which leads to the generation of ontology networks consisting of ontologies and mappings between these ontologies. However, neither developing ontologies nor finding mappings between ontologies is an easy task. It may happen that the ontologies are not consistent or complete, or the mappings between these ontologies are not correct or complete, or the resulting ontology network is not consistent. This may lead to problems when they are used in semantically-enabled applications. In this thesis, we address two issues relevant to the quality of the mappings and the structure in the ontology network. The first issue deals with the missing mappings between networked ontologies. Assuming existing mappings between ontologies are correct, we investigate whether and how to use these existing mappings, to find more mappings between ontologies. We propose and test several strategies of using the given correct mappings to align ontologies. The second issue deals with the missing structure, in particular missing is-a relations, in networked ontologies. Based on the assumption that missing is-a relations are a kind of modeling defects, we propose an ontology debugging approach to tackle this issue. We develop an algorithm for detecting missing is-a relations in ontologies, as well as algorithms which assist the user in repairing by generating and recommending possible ways of repairing and executing the repairing. Based on this approach, we develop a system and test its use and performance.

Ontologies are widely used in various domains, including biomedical research, to structure information, represent knowledge, and analyze data. The combination of ontologies from different domains is crucial for systematic data analysis and comparison of similar domains. This process requires ontology composition, integration, and alignment, which involve creating new classes by reusing classes from different domains, aggregating types of ontologies within the same domain, and finding correspondences between ontologies within the same or similar domain. This thesis presents use cases where we applied ontology composition, integration, and alignment of phenotype ontologies, and evaluated the resulting ontologies and alignment. First, we analyzed a large aging dataset of inbred laboratory mice, using Mouse Anatomy and Mouse Pathology ontologies. Second, we integrated phenotype ontologies for human and model organism phenotypes to enable comparisons of phenotypes between and within individual species. We developed Pheno-e, an extension of PhenomeNet. We identified novel abnormal anatomical classes for fly phenotypes, allowing the annotation of fly genes that were not annotated before. We demonstrate the distinct contributions of each species' phenotypic data to detecting human diseases using Pheno-e, and show that mouse phenotypic data contributes the most to the discovery of gene--disease associations. This work could guide the selection of model organisms when building methods to find gene-disease associations. Additionally, we refined class definitions in phenotypic ontologies, specifically targeting cell cardinality phenotypes. This representation resolved incorrect inferences in the utilized ontologies, enabling accurate interpretation of phenotypic descriptions. Our findings reveal that this correction enhances gene-disease prediction for diseases associated with cardinality phenotypes. Third, we introduce a novel neural-symbolic method that combines logic fundamentals with machine learning for ontology alignment. This method begins with symbolic representation, followed by iterative neural learning for alignment and symbolic representation consistency checking and reasoning, and back to neural learning. We demonstrate that our system generates noncontroversial alignments first and these alignments are coherent with respect to OWL EL. This novel method can pave the way for more accurate and efficient ontology-based methods, which can have significant implications for various semantic web applications.

No description available.

No description available.

[pt] Ontologias são vistas como uma solução para o problema de interoperabilidade em muitas aplicações como integração de banco de dados, comércio eletrônico, serviços web e redes sociais. Contudo, ao utilizar diferentes ontologias, persiste o problema de incompatibilidade semântica entre os modelos. Dentre os desafios para o campo de alinhamento de ontologia, está a seleção de alinhadores e a auto-configuração desses alinhadores . Este trabalho tem por objetivo tratar o problema de meta-alinhamento de ontologias utilizando métodos heurísticos para calibragem de alinhadores escolhidos sem detrimento do desempenho do sistema. Para avaliar esta proposta em comparação a outras propostas da literatura, optou-se por utilizar um benchmark para alinhamentos de ontologias amplamente utilizado por pesquisadores desse campo. A comparação entre as propostas é realizada com base em medidas de qualidade de alinhamentos sugeridas pelo benchmark. Além da avaliação quantitativa proposta pelo benchmark, é realizada uma comparação qualitativa em relação às características desejáveis de meta-alinhadores propostas na literatura. / [en] Ontologies are seen as a solution for many applications such as database integration, e-commerce, web services and social networks. However, the use of distinet ontologies does not solve the semantic interoperability problem among modls. Matcher selection and self-configuring are challenges from the ontology matching field. This work aims to treat the ontology metamatching problem using heuristic methods to fast tune a set of matchers. To evalutate this proporsal, a well-known benchark for ontology matching is used. The comparison among meta-matching approaches includes the benchmark evaluation metries and qualitative metries for meta-matching proposed in the literature.

Représentation et raisonnement qualitatifs fonctionnent avec des relations non-numériques entre les objets d'un univers. Les formalismes généraux développés dans ce domaine sont basés sur différents types d'algèbres de relations, comme les algèbres de Tarski. Tous ces formalismes, qui sont appelés des calculs qualitatifs, partagent l'hypothèse implicite que l'univers est homogène, c'est-à-dire qu'il se compose d'objets de même nature. Toutefois, les objets de différents types peuvent aussi entretenir des relations. L'état de l'art du raisonnement qualitatif ne permet pas de combiner les calculs qualitatifs pour les différents types d'objets en un seul calcul.De nombreuses applications discriminent entre différents types d'objets. Par exemple, certains modèles spatiaux discriminent entre les régions, les lignes et les points, et différentes relations sont utilisées pour chaque type d'objets. Dans l'alignement d'ontologies, les calculs qualitatifs sont utiles pour exprimer des alignements entre un seul type d'entités, telles que des concepts ou des individus. Cependant, les relations entre les individus et les concepts, qui imposent des contraintes supplémentaires, ne sont pas exploitées.Cette thèse introduit la modularité dans les calculs qualitatifs et fournit une méthodologie pour la modélisation de calculs qualitatifs des univers hétérogènes. Notre contribution principale est un cadre basé sur une classe spéciale de schémas de partition que nous appelons modulaires. Pour un calcul qualitatif engendré par un schéma de partition modulaire, nous définissons une structure qui associe chaque symbole de relation avec un domaine et codomain abstrait à partir d'un treillis booléen de sortes. Un module d'un tel calcul qualitatif est un sous-calcul limité à une sorte donnée, qui est obtenu par une opération appelée relativisation à une sorte. D'un intérêt pratique plus grand est l'opération inverse, qui permet de combiner plusieurs calculs qualitatifs en un seul calcul. Nous définissons une opération appelée combinaison modulo liaison, qui combine deux ou plusieurs calculs qualitatifs sur différents univers, en fonction de quelques relations de liaison entre ces univers. Le cadre est suffisamment général pour soutenir la plupart des calculs spatio-temporels qualitatifs connus. / Qualitative representation and reasoning operate with non-numerical relations holding between objects of some universe. The general formalisms developed in this field are based on various kinds of algebras of relations, such as Tarskian relation algebras. All these formalisms, which are called qualitative calculi, share an implicit assumption that the universe is homogeneous, i.e., consists of objects of the same kind. However, objects of different kinds may also entertain relations. The state of the art of qualitative reasoning does not offer a combination operation of qualitative calculi for different kinds of objects into a single calculus.Many applications discriminate between different kinds of objects. For example, some spatial models discriminate between regions, lines and points, and different relations are used for each kind of objects. In ontology matching, qualitative calculi were shown useful for expressing alignments between only one kind of entities, such as concepts or individuals. However, relations between individuals and concepts, which impose additional constraints, are not exploited.This dissertation introduces modularity in qualitative calculi and provides a methodology for modeling qualitative calculi with heterogeneous universes. Our central contribution is a framework based on a special class of partition schemes which we call modular. For a qualitative calculus generated by a modular partition scheme, we define a structure that associates each relation symbol with an abstract domain and codomain from a Boolean lattice of sorts. A module of such a qualitative calculus is a sub-calculus restricted to a given sort, which is obtained through an operation called relativization to a sort. Of a greater practical interest is the opposite operation, which allows for combining several qualitative calculi into a single calculus. We define an operation called combination modulo glue, which combines two or more qualitative calculi over different universes, provided some glue relations between these universes. The framework is general enough to support most known qualitative spatio-temporal calculi.

[pt] A abordagem de Dados Abertos tem como objetivo promover a interoperabilidade de dados na Web. Consiste na publicação de informações em formatos que permitam seu compartilhamento, descoberta, manipulação e acesso por parte de usuários e outros aplicativos de software. Essa abordagem requer a triplificação de conjuntos de dados, ou seja, a conversão do esquema de bases de dados relacionais, bem como suas instâncias, em triplas RDF. Uma questão fundamental neste processo é decidir a forma de representar conceitos de esquema de banco de dados em termos de classes e propriedades RDF. Isto é realizado através do mapeamento das entidades e relacionamentos para um ou mais vocabulários RDF, usados como base para a geração das triplas. A construção destes vocabulários é extremamente importante, porque quanto mais padrões são utilizados, melhor o grau de interoperabilidade com outros conjuntos de dados. No entanto, as ferramentas disponíveis atualmente não oferecem suporte adequado ao reuso de vocabulários RDF padrão no processo de triplificação. Neste trabalho, apresentamos o processo StdTrip, que guia usuários no processo de triplificação, promovendo o reuso de vocabulários de forma a assegurar interoperabilidade dentro do espaço da Linked Open Data (LOD). / [en] Open Data is a new approach to promote interoperability of data in the Web. It consists in the publication of information produced, archived and distributed by organizations in formats that allow it to be shared, discovered, accessed and easily manipulated by third party consumers. This approach requires the triplification of datasets, i.e., the conversion of database schemata and their instances to a set of RDF triples. A key issue in this process is deciding how to represent database schema concepts in terms of RDF classes and properties. This is done by mapping database concepts to an RDF vocabulary, used as the base for generating the triples. The construction of this vocabulary is extremely important, because the more standards are reused, the easier it will be to interlink the result to other existing datasets. However, tools available today do not support reuse of standard vocabularies in the triplification process, but rather create new vocabularies. In this thesis, we present the StdTrip process that guides users in the triplification process, while promoting the reuse of standard, RDF vocabularies.

Cette thèse propose l’utilisation de l’alignement d'ontologies pour contribuer à l'interopérabilité d’une fédération d’entreprises en se basant sur l'interopérabilité des données au niveau sémantique. Une approche d'alignement basée sur des modèles d’ontologie utilisant les mots noyaux est proposée en réponse aux problèmes et aux défis existants, visant ainsi à améliorer la capacité d'adaptation et la précision dans la mise en correspondance de concepts. De plus une étape d'agrégation des « matchers » analytique, qui permet de combiner automatiquement plusieurs adaptateurs et d'améliorer les résultats combinés, vient compléter l’approche. Un système prototype a été mis en œuvre à l’issue des travaux conceptuels pour la validation de l’approche proposée. Les expériences démontrent que l’approche proposée a obtenu des résultats prometteurs et a atteint les objectifs escomptés sur la définition de proximité des concepts. L'approche d'alignement d’ontologies proposée et le système de prototype mis en œuvre ont enfin été appliqués à une architecture dirigée par les ontologies et axée sur l'interrogation des données de plusieurs bases de données relationnelles. / This thesis brings ontology alignment to contribute to federated enterprise interoperability focusing on data interoperability at the semantic level. In response to existing problems and challenges, aiming at improving the matching ability and precision, a pattern-based core word ontology alignment approach is proposed, as well as an analytic matcher aggregation approach, which allows combining the multiple matchers automatically and improve the combined results. A prototype system is implemented for validation and further application based on the proposed approaches. The experiments suggest that the proposed approaches obtained promising results and reached expected goals. The proposed ontology alignment approach and implemented prototype system are applied to an ontology-driven architecture for querying data from multiple relational databases.

Dans cette thèse, nous présentons des approches d’adaptation d’un processus d’alignement aux caractéristiques des ontologies alignées, qu'il s'agisse de caractéristiques quantitatives telles que leur volume ou de caractéristiques particulières liées par exemple à la façon dont les labels des concepts sont construits. Concernant les caractéristiques quantitatives, nous proposons deux méthodes de partitionnement d'ontologies qui permettent l’alignement des ontologies très volumineuses. Ces deux méthodes génèrent, en entrée du processus d'alignement, des sous ensembles de taille raisonnable des deux ontologies à aligner en prenant en compte dès le départ l'objectif d'alignement dans le processus de partitionnement.Concernant les caractéristiques particulières des ontologies alignées, nous présentons l’environnement TaxoMap Framework qui permet la spécification de traitements de raffinement à partir de primitives prédéfinies. Nous proposons un langage de patrons MPL (the Mapping Pattern Language) que nous utilisons pour spécifier les traitements de raffinement.En plus des approches d’adaptation aux caractéristiques des ontologies alignées, nous présentons des approches de réutilisation des résultats d'alignement pour l'ingénierie ontologique. Nous nous focalisons plus particulièrement sur l'utilisation de l'alignement pour l'enrichissement d'ontologies. Nous étudions l'apport des techniques d'alignement pour l'enrichissement et l'impact des caractéristiques de la ressource externe utilisée comme source d'enrichissement. Enfin, nous présentons la façon dont l'environnement TaxoMap Framework a été implémenté et les expérimentations réalisées : des tests sur le module d'alignement TaxoMap, sur l'approche de raffinement de mappings, sur les méthodes de partitionnement d'ontologies de très grande taille et sur l'approche d'enrichissement d'ontologies. / In this thesis, we present approaches for adapting an alignment process to the characteristics of the aligned ontologies, whether in respect of the quantitative characteristics such as their volume or the particular characteristics related for example to the way in which the labels of the concepts are built.Concerning the quantitative characteristics, we propose two ontology partitioning methods that make the alignment of very large ontologies possible. Both methods generate in the input of the alignment process, subsets of reasonable size of the two ontologies to be aligned by taking into account, as soon as possible, the alignment objective in the partitioning process.Concerning the particular characteristics of the aligned ontologies, we present the TaxoMap Framework environment that allows the specification of refinement treatments from predefined primitives. We propose a pattern language MPL (the Mapping Pattern Language) that we use to specify the refinement treatments.In addition to the adaptation approaches to the characteristics of the aligned ontologies, we present approaches for re-using the alignment results for the ontological engineering. We focus specifically on the use of the alignment for the ontology enrichment. We study the contribution of the alignment techniques for the enrichment and the impact of the characteristics of the external resource used as a source of enrichment.Finally, we present how the TaxoMap Framework environment was implemented and the performed experiments: tests on the TaxoMap alignment module, on the mappings refinement approach, on the partitioning methods of very large ontologies and on the ontology enrichment approach.

[ES] La estimación del trasvase de votos de unos partidos a otros entre dos procesos electorales es un tema relevante dentro de la sociología política. Dadas las limitaciones de las encuestas en este campo, existe un interés creciente en los procedimientos de estimación que utilizan exclusivamente la realidad incuestionable de los propios resultados electorales. lphom es uno de estos métodos. lphom es un procedimiento, relativamente sencillo, que, tras ser aplicado en estos últimos años en algunos estudios reales, ha proporcionado, en todos los casos, resultados muy razonables. En la presente tesis se aborda el problema, hasta ahora no investigado, de desarrollar procedimientos para estimar los márgenes de incertidumbre de los resultados obtenidos al aplicar este procedimiento en estudios reales. Tras definir unos indicadores adecuados, EI y EP, para cuantificar dicha incertidumbre, el proceso se basa en utilizar los residuos del modelo lphom para (i) evaluar el grado de incumplimiento de la hipótesis de homogeneidad en la que se sustenta lphom, y (ii) investigar, mediante un conjunto de estudios de simulación desarrollados en diferentes escenarios, las relaciones entre estos indicadores. Las conclusiones generales de estos estudios permiten desarrollar unos procedimientos para estimar los valores previsibles de EI y EP en estudios reales, así como los límites de confianza superior para los mismos. También se desarrolla un procedimiento para obtener intervalos de confianza para las diferentes probabilidades de transición de votos pjk estimadas mediante lphom. Los métodos desarrollados se ilustran aplicándolos a la estimación del trasvase de votos entre la primera y la segunda vuelta de las elecciones presidenciales francesas de 2017. Se indican finalmente una serie de líneas futuras de investigación sugeridas por los resultados de esta tesis. / [CAT] L'estimació del transvasament de vots d'uns partits a altres entre dos processos electorals és un tema rellevant dins de la sociología política. Donades les limitacions de les enquestes en aquest camp, hi ha un creixent interés en els procediments d'estimació que utilitzen exclusivamente la realitat inqüestionable dels propis resultats electorals. En aquest sentit, lphom és un mètode senzill que després de ser aplicat aquestos últims anys en alguns estudis reals, ha proporcionat en tots els casos resultats molt razonables. En la següent tesi, s'aborda el problema, encara no investigat, de desenvolupar procediments per a estimar els margens d'incentidumbre dels resultats obtenguts al aplicar aquest procediment en els estudis reals. Després d'haber definit uns indicadors adecuats, EI y EP, per a quantificar dita incertidumbre, el procés es basa en utilitzar els residus del model lphom per a (i) evaluar el grau de incompliment de la hipótesis d'homogeneïtat en què es sustenta lphom, i (ii) investigar, mitjançant un conjunt d'estudis de simulacions desenvolupades en diferents escenaris, les relacions entre aquestos indicadors. Les conclusions generals d'aquestos estudis permeten desenvolupar uns procediments per a estimar els valors previsibles de EI y EP en estudis reals, així com els límits de confianza superior per als mateixos. També s'aconsegueix un procediment per a obtindre intervals de confiança per a les diferents probabilitats de transició de vots pjk estimats mediant lphom. Els métodes desenvolupats s'ilustren aplicats a l'estimació del transvasament de vots entre la primera y segona volta de les eleccions presidencials franceses de 2017. S'indiquen finalment una sèrie de línies futures d'investigació suggerides pels resultats d'aquesta tesi. / [EN] The estimation of the transfer of votes from some political parties to another between two electoral processes is a relevant issue within political sociology. Given the limitations of the surveys in this field, there is a growing interest in estimation procedures that exclusively use the unquestionable reality of the electoral results themselves. In this sense, lphom is a simple method that, after being applied in recent years in some real studies, has provided, in all cases, very reasonable results. This thesis addresses the problem, so far not investigated, of developing procedures to estimate the margins of uncertainty of the results obtained by applying this procedure in real studies. After defining suitable indicators, EI and EP, to quantify this uncertainty, the process is based on using the residuals of the lphom model to (i) evaluate the degree of non-compliance with the homogeneity hypothesis on which lphom is based, and (ii) investigate, through a set of simulation studies developed in different scenarios, the relationships between these indicators. The general conclusions of these studies allow the development of procedures to estimate the predictable values of IE and PE in real studies, as well as the upper confidence limits for them. A procedure is also developed to obtain confidence intervals for the different transition probabilities of votes pjk estimated by lphom. The methods developed are illustrated by applying them to the estimation of the transfer of votes between the first and second rounds of the 2017 French presidential elections. A series of future lines of research suggested by the results of this thesis are finally indicated. / Martín Marín, J. (2020). Análisis de la incertidumbre en la estimación de la movilidad electoral mediante el procedimiento LPHOM [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/157638