xiii, 149 p. : ill. (some col.) / Information extraction (IE) aims to recognize and retrieve certain types of information from natural language text. For instance, an information extraction system may extract key geopolitical indicators about countries from a set of web pages while ignoring other types of information. IE has existed as a research field for a few decades, and ontology-based information extraction (OBIE) has recently emerged as one of its subfields. Here, the general idea is to use ontologies--which provide formal and explicit specifications of shared conceptualizations--to guide the information extraction process. This dissertation presents two novel directions for ontology-based information extraction in which ontologies are used to improve the information extraction process. First, I describe how a component-based approach for information extraction can be designed through the use of ontologies in information extraction. A key idea in this approach is identifying components of information extraction systems which make extractions with respect to specific ontological concepts. These components are termed "information extractors". The component-based approach explores how information extractors as well as other types of components can be used in developing information extraction systems. This approach has the potential to make a significant contribution towards the widespread usage and commercialization of information extraction. Second, I describe how an ontology-based information extraction system can make use of multiple ontologies. Almost all previous systems use a single ontology, although multiple ontologies are available for most domains. Using multiple ontologies in information extraction has the potential to extract more information from text and thus leads to an improvement in performance measures. The concept of information extractor, conceived in the component-based approach for information extraction, is used in designing the principles for accommodating multiple ontologies in an ontology-based information extraction system. / Committee in charge: Dr. Dejing Dou, Chair; Dr. Arthur Farley, Member; Dr. Michal Young, Member; Dr. Monte Westerfield, Outside Member

Ces dernières années de nombreux progrès ont été faits dans le domaine de l'extraction de relations à partir de textes, facilitant ainsi la construction de ressources lexicales ou sémantiques. Cependant, les méthodes proposées (apprentissage supervisé, méthodes à noyaux, apprentissage distant, etc.) n'exploitent pas tout le potentiel des textes : elles ont généralement été appliquées à un niveau phrastique, sans tenir compte des éléments de mise en forme. Dans ce contexte, l'objectif de cette thèse est d'adapter ces méthodes à l'extraction de relations exprimées au-delà des frontières de la phrase. Pour cela, nous nous appuyons sur la sémantique véhiculée par les indices typographiques (puces, emphases, etc.) et dispositionnels (indentations visuelles, retours à la ligne, etc.), qui complètent des formulations strictement discursives. En particulier, nous étudions les structures énumératives verticales qui, bien qu'affichant des discontinuités entre leurs différents composants, présentent un tout sur le plan sémantique. Ces structures textuelles sont souvent révélatrices de relations hiérarchiques. Notre travail est divisé en deux parties. (i) La première partie décrit un modèle pour représenter la structure hiérarchique des documents. Ce modèle se positionne dans la suite des modèles théoriques proposés pour rendre compte de l'architecture textuelle : une abstraction de la mise en forme et une connexion forte avec la structure rhétorique sont faites. Toutefois, notre modèle se démarque par une perspective d'analyse automatique des textes. Nous en proposons une implémentation efficace sous la forme d'une méthode ascendante et nous l'évaluons sur un corpus de documents PDF.(ii) La seconde partie porte sur l'intégration de ce modèle dans le processus d'extraction de relations. Plus particulièrement, nous nous sommes focalisés sur les structures énumératives verticales. Un corpus a été annoté selon une typologie multi-dimensionnelle permettant de caractériser et de cibler les structures énumératives verticales porteuses de relations utiles à la création de ressources. Les observations faites en corpus ont conduit à procéder en deux étapes par apprentissage supervisé pour analyser ces structures : qualifier la relation puis en extraire les arguments. L'évaluation de cette méthode montre que l'exploitation de la mise en forme, combinée à un faisceau d'indices lexico-syntaxiques, améliore les résultats. / The past decade witnessed significant advances in the field of relation extraction from text, facilitating the building of lexical or semantic resources. However, the methods proposed so far (supervised learning, kernel methods, distant supervision, etc.) don't fully exploit the texts : they are usually applied at the sentential level and they don't take into account the layout and the formatting of texts. In such a context, this thesis aims at expanding those methods and makes them layout-aware for extracting relations expressed beyond sentence boundaries. For this purpose, we rely on the semantics conveyed by typographical (bullets, emphasis, etc.) and dispositional (visual indentations, carriage returns, etc.) features. Those features often substitute purely discursive formulations. In particular, the study reported here is dealing with the relations carried by the vertical enumerative structures. Although they display discontinuities between their various components, the enumerative structures can be dealt as a whole at the semantic level. They form textual structures prone to hierarchical relations. This study was divided into two parts. (i) The first part describes a model representing the hierarchical structure of documents. This model is falling within the theoretical framework representing the textual architecture : an abstraction of the layout and the formatting, as well as a strong connection with the rhetorical structure are achieved. However, our model focuses primarily on the efficiency of the analysis process rather than on the expressiveness of the representation. A bottom-up method intended for building automatically this model is presented and evaluated on a corpus of PDF documents. (ii) The second part aims at integrating this model into the process of relation extraction. In particular, we focused on vertical enumerative structures. A multidimensional typology intended for characterizing those structures was established and used into an annotation task. Thanks to corpus-based observations, we proposed a two-step method, by supervised learning, for qualifying the nature of the relation and identifying its arguments. The evaluation of our method showed that exploiting the formatting and the layout of documents, in combination with standard lexico-syntactic features, improves those two tasks.

Orientador: Edberto Ferneda / Banca: Maria José Vicentini Jorente / Banca: Luciana Maria Vieira Pöttker / Resumo: Um sistema de recuperação de informação é um elemento mediador entre um acervo documental e os usuários que buscam por documentos relevantes. Nesse contexto, as interfaces desempenham uma função importante: em um primeiro momento, auxiliando o usuário na tarefa de expressar a sua necessidade de informação por meio de uma expressão de busca e, em um segundo momento, fornecendo recursos para ajudá-lo a selecionar documentos relevantes dentre os resultados obtidos. A recuperação de informação é um processo linguístico cuja eficiência depende de coincidências terminológicas entre a expressão de busca do usuário e a representação dos documentos. Este trabalho propõe um modelo de interface na qual a estrutura terminológica de uma ontologia é utilizada para auxiliar o usuário na seleção de documentos relevantes dentre aqueles resultantes de sua busca. Caracteriza-se como uma pesquisa de natureza aplicada, e exploratória e bibliográfica quanto aos procedimentos. Conclui-se que a apresentação visual de uma ontologia permite o desenvolvimento de interfaces dinâmicas e interativas, proporcionando ao usuário uma navegação estimulante e prazerosa por entre os documentos resultantes de sua busca, tendo por base os termos de uma determinada área de conhecimento. / Abstract: An information retrieval system is a mediating element between a document collection and the users who looking for relevant documents. In this context, interfaces play an important role: firstly, assisting the user to expressing their information need by means of a search expression, and secondly by providing resources to help selecting relevant documents from the obtained results. The information retrieval is a linguistic process whose efficiency depends on terminological coincidences between the user's query and the representation of documents. This work proposes an interface model in which the terminological structure of an ontology is used to assist the user in the selection of relevant documents among those resulting from their search. It is characterized as an applied, exploratory and bibliographic research. It is concluded that the visual presentation of ontology allows the development of dynamic and interactive interfaces, providing the user with stimulating and pleasant navigation among the documents resulting from their search, based on the terms of a certain knowledge area. / Mestre

Individualmente, as linhas de pesquisa da Inteligência Artificial têm proposto abordagens para a resolução de inúmeros problemas complexos do mundo real. O paradigma orientado a agentes provê os agentes autônomos, capazes de perceber os seus ambientes, reagir de acordo com diferentes circunstâncias e estabelecer interações sociais com outros agentes de software ou humanos. As redes Bayesianas fornecem uma maneira de representar graficamente as distribuições de probabilidades condicionais e permitem a realização de raciocínios probabilísticos baseados em evidências. As ontologias são especificações explícitas e formais de conceituações, que são usadas em uma variedade de áreas de pesquisa, incluindo os Sistemas Multiagentes. Contudo, existem aplicações cujos requisitos não podem ser atendidos por uma única tecnologia. Circunstâncias como estas exigem a integração de tecnologias desenvolvidas por distintas áreas da Ciência da Computação. Esta dissertação trata a integração do modelo de agentes BDI (Belief-Desire-Intention) e das redes Bayesianas. Além disso, é adotada uma abordagem baseada em ontologias para representar o conhecimento incerto dos agentes. O primeiro passo em direção a integração foi o desenvolvimento de uma ontologia para representar a estrutura das redes Bayesinas. Esta ontologia tem como principal objetivo permitir a interoperabilidade agentes compatíveis com a arquitetura proposta. No entanto, a ontologia também facilita o entendimento necessário para abstrair os estados mentais e processos cognitivos dos agentes através de elementos das redes Bayesianas. Uma vez construída a ontologia, a mesma foi integrada com a arquitetura BDI. Através da integração do modelo BDI com as redes Bayesianas foi obtida uma arquitetura cognitiva de agentes capaz de deliberar sob incerteza. O processo de integração foi composto de duas etapas: abstração dos estados mentais através de elementos das redes Bayesianas e especificação do processo deliberativo. Finalmente, foi desenvolvido um estudo de caso, que consistiu na aplicação da arquitetura proposta no Agente Social, um componente de um portal educacional multiagente (PortEdu). / Individually, Artificial Intelligence research areas have proposed approaches to solve several complex real-world problems. The agent-based paradigm provided autonomous agents, capable of perceiving their environment, reacting in accordance with different situations, and establishing social interactions with other software agents and humans. Bayesian networks provided a way to represent graphically the conditional probability distributions and an evidence-based probabilistic reasoning. Ontologies are an effort to develop formal and explicit specifications of concepts, which have been used by a wide range of research areas, including Multiagent Systems. However, there are applications whose requirements can not be addressed by a single technology. Circumstances like these demand the integration of technologies developed by distinct areas of Computer Science. This work is particularly concerned with the integration of Belief-Desire-Intention (BDI) agent architecture and Bayesian networks. Moreover, it is adopted an ontology-based approach to represent the agent’s uncertain knowledge. To bring together those technologies, it was developed an ontology to represent the structure of Bayesian networks knowledge representation. This ontology supports the interoperability among agents that comply with the proposed architecture, and it also facilitates the understanding necessary to abstract the agents’ mental states and cognitive processes through elements of Bayesian networks. Once specified the ontology, it was integrated with the BDI agent architecture. By integrating BDI architecture and Bayesian networks, it was obtained a cognitive agent architecture capable of reasoning under uncertainty. It was performed in two stages: abstraction of mental states through Bayesian networks and specification of the deliberative process. Finally, it was developed a case study, which consists in applying the probabilistic BDI architecture in the Social Agent, a component of a multiagent educational portal (PortEdu).

Redes de Quarta Geração (4G) possibilitam acesso ubíquo a serviços em redes heterogêneas. Nesses ambientes, a distribuição do gerenciamento pode necessitar de semânticas para o vocabulário compartilhado, uma facilidade ainda não explorada. Por essa razão, provedores de acesso de diferentes domínios de gerenciamento possuem o desafio de compartilhar informações gerenciais nas trocas de pontos de acesso realizadas pelos usuários em redes 4G. Isso se deve principalmente à carência de um modelo de negócio com controle de acesso, com acordos mútuos, e com gerenciamento de tarifação, segurança e privacidade , o que limita a colaboração entre provedores no momento de aceitar um novo usuário. Este trabalho descreve um conjunto de ontologias para serem utilizadas por novos modelos de acesso com suporte de ontologias para redes 4G. As ontologias formalizam o vocabulário comum das entidades em redes 4G. Para avaliar as ontologias desenvolvidas, a abordagem escolhida foi desenvolver dois protótipos de serviços para redes 4G que especializaram partes das ontologias Haggle, Y-Comm, SOHand e DOHand. A principal contribuição deste trabalho é facilitar a troca de informações gerenciais entre moderadores, usuários e provedores de serviços. Outra contribuição é mostrar como o uso de ontologias em plataformas 4G facilita a inserção de novos serviços nos ambientes integrados / Fourth Generation Networks (4G) introduce the capability of ubiquitous access to services in heterogeneous networks. In these environments, the distribution of the management would require semantics for common sharing vocabulary, an ability which is not yet provided. For that reason, service providers from different management domains are facing the challenge of exchanging managerial information between them during handovers. This is mainly due to the lack of new business models with access control, peering agreements, billing, security and privacy management, QoS standardization which strongly limit the collaboration between providers in the moment of receiving a new user. This work describes a set of ontologies to be used by new access models supported by ontologies for 4G. The ontologies are mapping out the common vocabulary of entities. The approach used to validate the set of ontologies was the development of two service prototypes for 4G, the prototypes specialized a subset of Haggle, Y-Comm, SOHand and DOHand ontologies. The main contribution is to provide the information structuring necessary for easy exchanging of managerial information between service providers, users and brokers. Another contribution of this work is to demostrate how ontologies facilitate the creation of new services in integrated environments

Individualmente, as linhas de pesquisa da Inteligência Artificial têm proposto abordagens para a resolução de inúmeros problemas complexos do mundo real. O paradigma orientado a agentes provê os agentes autônomos, capazes de perceber os seus ambientes, reagir de acordo com diferentes circunstâncias e estabelecer interações sociais com outros agentes de software ou humanos. As redes Bayesianas fornecem uma maneira de representar graficamente as distribuições de probabilidades condicionais e permitem a realização de raciocínios probabilísticos baseados em evidências. As ontologias são especificações explícitas e formais de conceituações, que são usadas em uma variedade de áreas de pesquisa, incluindo os Sistemas Multiagentes. Contudo, existem aplicações cujos requisitos não podem ser atendidos por uma única tecnologia. Circunstâncias como estas exigem a integração de tecnologias desenvolvidas por distintas áreas da Ciência da Computação. Esta dissertação trata a integração do modelo de agentes BDI (Belief-Desire-Intention) e das redes Bayesianas. Além disso, é adotada uma abordagem baseada em ontologias para representar o conhecimento incerto dos agentes. O primeiro passo em direção a integração foi o desenvolvimento de uma ontologia para representar a estrutura das redes Bayesinas. Esta ontologia tem como principal objetivo permitir a interoperabilidade agentes compatíveis com a arquitetura proposta. No entanto, a ontologia também facilita o entendimento necessário para abstrair os estados mentais e processos cognitivos dos agentes através de elementos das redes Bayesianas. Uma vez construída a ontologia, a mesma foi integrada com a arquitetura BDI. Através da integração do modelo BDI com as redes Bayesianas foi obtida uma arquitetura cognitiva de agentes capaz de deliberar sob incerteza. O processo de integração foi composto de duas etapas: abstração dos estados mentais através de elementos das redes Bayesianas e especificação do processo deliberativo. Finalmente, foi desenvolvido um estudo de caso, que consistiu na aplicação da arquitetura proposta no Agente Social, um componente de um portal educacional multiagente (PortEdu). / Individually, Artificial Intelligence research areas have proposed approaches to solve several complex real-world problems. The agent-based paradigm provided autonomous agents, capable of perceiving their environment, reacting in accordance with different situations, and establishing social interactions with other software agents and humans. Bayesian networks provided a way to represent graphically the conditional probability distributions and an evidence-based probabilistic reasoning. Ontologies are an effort to develop formal and explicit specifications of concepts, which have been used by a wide range of research areas, including Multiagent Systems. However, there are applications whose requirements can not be addressed by a single technology. Circumstances like these demand the integration of technologies developed by distinct areas of Computer Science. This work is particularly concerned with the integration of Belief-Desire-Intention (BDI) agent architecture and Bayesian networks. Moreover, it is adopted an ontology-based approach to represent the agent’s uncertain knowledge. To bring together those technologies, it was developed an ontology to represent the structure of Bayesian networks knowledge representation. This ontology supports the interoperability among agents that comply with the proposed architecture, and it also facilitates the understanding necessary to abstract the agents’ mental states and cognitive processes through elements of Bayesian networks. Once specified the ontology, it was integrated with the BDI agent architecture. By integrating BDI architecture and Bayesian networks, it was obtained a cognitive agent architecture capable of reasoning under uncertainty. It was performed in two stages: abstraction of mental states through Bayesian networks and specification of the deliberative process. Finally, it was developed a case study, which consists in applying the probabilistic BDI architecture in the Social Agent, a component of a multiagent educational portal (PortEdu).

Made available in DSpace on 2018-06-27T18:56:06Z (GMT). No. of bitstreams: 3 Heleno Carmo Borges Cabral.pdf: 5857862 bytes, checksum: d2951b87950c843d49760204923bdc2a (MD5) Heleno Carmo Borges Cabral.pdf.txt: 118091 bytes, checksum: e4e222642f43d380a5e6bf3a2acd2aac (MD5) Heleno Carmo Borges Cabral.pdf.jpg: 3434 bytes, checksum: 3c397ca213b4291d6533cce010f30ed4 (MD5) Previous issue date: 2010-06-23 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Ontocancro is an ontology stored in a knowledge database designed to be a source of information to integrate transcriptomics and interatomics data involved in gene pathways of genome maintenance/stability mechanisms (GMM). Genome maintenance mechanisms are shown to be critical for cell homeostasis since their malfunctioning can predispose to cancer. Repair, apoptosis and chromosome stability pathways comprise the cornerstone of GMM. The information about these pathways are disseminated in various databases as NCI-Nature, BioCarta, KEGG, Reactome, Prosite, GO and others. Ontocancro was created with the intention of integratin the information of genes involved in GMM from several curated databases. This data integration is difficult for biological data lack a unified vocabulary and need constant update what is provided by Ontocancro. Additionally, it allows the integration of transcriptome data provided by some Affymetrix microarrays platforms with interactome data from the STRING database, which has information about protein interactions. So, this work shows the integration of data from biological information systems using the ontology paradigm, in order to integrate transcriptomics and interatomics data involved in gene pathways of genome stability. / A Ontocancro é uma ontologia armazenada em um banco de dados de conhecimento projetada para ser a fonte de informação referente a integração de dados de interatoma e transcriptoma envolvidos em vias metabólicas de mecanismo de manutenção do genoma humano (GMM). Esse mecanismo de manutenção são críticos para homeostase celular desde o seu mau funcionamento, o que pode causar câncer. O reparo, a apoptose e as vias de estabilidade cromossômicas compreendem o cerne do GMM. A informação sobre essas vias metabólicas são disseminadas em vários bancos de dados, como o NCI-Nature, o BioCarta, o KEGG, o Reactome, o Prosite e o GO, entre outros. A ontologia Ontocancro foi criada com a intenção de integrar a informação sobre os genes envolvidos em GMM a partir de diversos bancos de dados curados. Essa integração de dados é complexa pela falta de um vocabulário sobre os dados biológicos e a necessidade constante de atualização destes dados. Para sanar essas duas dificuldades, a Ontocancro foi criada. Adicionalmente, ela permite a integração de dados oriundos de transcriptoma obtidos a partir da plataforma Affymetrix com os dados de interatoma obtidos a partir do banco de dados chamado STRING, o qual possui informação sobre as interações entre as proteínas. Portanto, este trabalho apresenta a integração de dados obtidos de sistemas de informação biológicos usando o paradigma ontológico, de forma a integrar os dados envolvidos em interatoma e transcriptoma em vias metabólicas de estabilidade do genoma.

Atualmente uma grande quantidade de conteúdo é produzida e publicada todos os dias na Internet. São documentos publicados por diferentes pessoas, por diversas organizações e em inúmeros formatos sem qualquer tipo de padronização. Por esse motivo, a informação relevante sobre um mesmo domínio de interesse acaba espalhada pela Web nos diversos portais, o que dificulta uma visão ampla, centralizada e objetiva sobre esta informação. Nesse contexto, a integração dos dados espalhados na rede torna-se um problema de pesquisa relevante, para permitir a realização de consultas mais inteligentes, de modo a obter resultados mais ricos de significado e mais próximos do interesse do usuário. No entanto, tal integração não é trivial, sendo por muitas vezes custosa devido à dependência do desenvolvimento de sistemas e mão de obra especializados, visto que são poucos os modelos reaproveitáveis e facilmente integráveis entre si. Assim, a existência de um modelo padronizado para a integração dos dados e para o acesso à informação produzida por essas diferentes entidades reduz o esforço na construção de sistemas específicos. Neste trabalho é proposta uma arquitetura baseada em ontologias para a integração de dados publicados na Internet. O seu uso é ilustrado através de casos de uso reais para a integração da informação na Internet, evidenciando como o uso de ontologias pode trazer resultados mais relevantes. / A lot of content is produced and published every day on the Internet. Those documents are published by different people, organizations and in many formats without any type of established standards. For this reason, relevant information about a domain of interest is spread through the Web in various portals, which hinders a broad, centralized and objective view of this information. In this context, the integration of the data scattered in the network becomes a relevant research problem, to enable smarter queries, in order to obtain richer results of meaning and closer to the user\'s interest. However, such integration is not trivial, and is often costly because of the reliance on the development of specialized systems by professionals, since there are few reusable and easily integrable models. Thus, the existence of a standardized model for data integration and access to the information produced by these different entities reduces the effort in the construction of specific systems. In this work we propose an architecture based on ontologies for the integration of data published on the Internet. Its use is illustrated through experimental cases for the integration of information on the Internet, showing how the use of ontologies can bring more relevant results.

Un mapping d’ontologies est un ensemble de correspondances. Chaque correspondance relie des artefacts, typiquement concepts et propriétés, d’une ontologie avec ceux d’une autre ontologie. Le mapping entre ontologies a suscité beaucoup d’intérêt durant ces dernières années. En effet, le mapping d’ontologies est largement utilisé pour mettre en oeuvre de l’interopérabilité et intégration (transformation de données, réponse à la requête, composition de web service) dans les applications, et également dans la création de nouvelles ontologies. D’une part, vérifier l’exactitude (logique) d’un mapping est devenu un prérequis fondamentale à son utilisation. D’autre part, pour deux ontologies données, plusieurs mappings peuvent être établis, obtenus par différentes méthodes d’alignement, ou définis manuellement. L’utilisation de plusieurs mappings entre deux ontologies dans une seule application ou pour synthétiser un seul mapping tirant profit de ces plusieurs mappings, peut générer des erreurs dans l’application ou dans le mapping synthétisé car ces plusieurs mappings peuvent être contradictoires. Dans les deux situations décrites ci-dessus, l’exactitude, la non-contradiction et autres propriétés sont généralement exprimées de façon formelle et vérifiées dans le contexte des ontologies formelles (par exemple, lorsque les ontologies sont représentées en logique) La vérification de ces propriétés est généralement effectuée à l’aide d’un seul formalisme, exigeant d’une part que les ontologies soient représentées par ce seul formalisme et, d’autre part, qu’une représentation formelle des mappings soit fournie, complétée par des notions formalisant les propriétés recherchées. Cependant, il existe une multitude de formalismes hétérogènes pour exprimer les ontologies, allant des plus informels (par exemple, du texte contrôlé, des modèles en UML) aux formels (par exemple, des logiques de description ou des catégories). Ceci implique que pour appliquer les approches existantes, les ontologies hétérogènes doivent être traduites (ou juste transformées, si l’ontologie source est exprimée de façon informelle ou si la traduction complète pour maintenir l’équivalence n’est pas possible) dans un seul formalisme commun et les mappings sont reformulés à chaque fois : seulement à l’issu de ce processus, les propriétés recherchées peuvent être établies. Même si cela est possible, ce processus peut produire à la fois des mappings corrects et incorrects vis-à-vis de ces propriétés, en fonction de la traduction (transformation) opérée. En effet, les propriétés recherchées dépendent du formalisme employé pour exprimer les ontologies et les mappings. Dans cette dissertation, des différentes propriétés ont été a été reformulées d’une manière unifiée dans le contexte d’ontologies hétérogènes utilisant la théorie de Galois. Dans ce contexte, les ontologies sont représentées comme treillis, et les mappings sont reformulés comme fonctions entre ces treillis. Les treillis sont des structures naturelles pour la représentation directe d’ontologies sans obligation de traduire ou transformer les formalismes dans lesquels les ontologies sont exprimées à l’origine. Cette reformulation unifiée a permis d’introduire une nouvelle notion de mappings compatibles et incompatibles. Il est ensuite formellement démontré que cette nouvelle notion couvre plusieurs parmi les propriétés recherchées de mappings, mentionnées dans l’état de l’art. L’utilisation directe de mappings compatibles et incompatibles est démontrée par l’application à des mappings d’ontologies de haut niveau. La notion de mappings compatibles et incompatibles est aussi appliquée sur des ontologies de domaine, mettant en évidence comment les mappings incompatibles génèrent des résultats incorrects pour la fusion d’ontologies. / An ontology mapping is a set of correspondences. Each correspondence relates artifacts, such as concepts and properties, of one ontology to artifacts of another ontology. In the last few years, a lot of attention has been paid to establish mappings between source ontologies. Ontology mapping is widely and effectively used for interoperability and integration tasks (data transformation, query answering, or web-service composition, to name a few), and in the creation of new ontologies. On the one side, checking the (logical) correctness of ontology mappings has become a fundamental prerequisite of their use. On the other side, given two ontologies, there are several ontology mappings between them that can be obtained by using different ontology matching methods or just stated manually. Using ontology mappings between two ontologies in combination within a single application or for synthesizing one mapping taking the advantage of two original mappings, may cause errors in the application or in the synthesized mapping because those original mappings may be contradictory (conflicting). In both situations, correctness is usually formalized and verified in the context of fully formalized ontologies (e.g. in logics), even if some “weak” notions of correctness have been proposed when ontologies are informally represented or represented in formalisms preventing a formalization of correctness (such as UML). Verifying correctness is usually performed within one single formalism, requiring on the one side that ontologies need to be represented in this unique formalism and, on the other side, a formal representation of mapping is provided, equipped with notions related to correctness (such as consistency). In practice, there exist several heterogeneous formalisms for expressing ontologies, ranging from informal (text, UML and others) to formal (logical and algebraic). This implies that, willing to apply existing approaches, heterogeneous ontologies should be translated (or just transformed if, the original ontology is informally represented or when full translation, keeping equivalence, is not possible) in one common formalism, mappings need each time to be reformulated, and then correctness can be established. This is possible but possibly leading to correct mappings under one translation and incorrect mapping under another translation. Indeed, correctness (e.g. consistency) depends on the underlying employed formalism in which ontologies and mappings are expressed. Different interpretations of correctness are available within the formal or even informal approaches questioning about what correctness is indeed. In the dissertation, correctness has been reformulated in the context of heterogeneous ontologies by using the theory of Galois connections. Specifically ontologies are represented as lattices and mappings as functions between those lattices. Lattices are natural structures for directly representing ontologies, without changing the original formalisms in which ontologies are expressed. As a consequence, the (unified) notion of correctness has been reformulated by using Galois connection condition, leading to the new notion of compatible and incompatible mappings. It is formally shown that the new notion covers the reviewed correctness notions, provided in distinct state of the art formalisms, and, at the same time, can naturally cover heterogeneous ontologies. The usage of the proposed unified approach is demonstrated by applying it to upper ontology mappings. Notion of compatible and incompatible ontology mappings is also applied on domain ontologies to highlight that incompatible ontology mappings give incorrect results when used for ontology merging.

Ontologies are a key technology for the Semantic Web. In different areas, a large number of ontologies have been developed so far by different people or organizations under the same domains and many of them contain overlapping information. In order to get more benefit from different ontologies having inter-related knowledge they have to be aligned or merged. A number of systems have been developed for aligning and merging ontologies and various alignment strategies are used in these systems. However, there is no system available which supports multiple alignment sessions for aligning large ontologies adequately. In this thesis work we propose a session-based framework for aligning and merging large ontologies. We have implemented two types of sessions, computation sessions to generate suggestions and validation sessions to validate these generated suggestions. Furthermore after categorizing suggestions into accepted and rejected ones, we generated partial reference alignment (PRA) that can be used to compute similarities between terms and to filter mapping suggestions. We have also proposed recommendation process integrated with computation and validation sessions in order to find out which matchers, and combinations are better to use for alignment process. Either computation and validation sessions may use the recommended settings or the user can select other matchers and combinations.