The semantic web is an extension of the current web that provides a standardstructure for data representation and reasoning, allowing content to be readable for both humans and machines in a form known as ontological knowledgebases. The goal of the Semantic Web is to be used in large-scale technologies or systems such as search engines, healthcare systems, and social mediaplatforms. Some challenges may deter further progress in the development ofthe Semantic Web and the associated web mining processes. In this reviewpaper, an overview of Semantic Web mining will examine and analyze challenges with data integration, dynamic knowledge-based methods, efficiencies,and data mining algorithms regarding ontological approaches. Then, a reviewof recent solutions to these challenges such as clustering, classification, association rule mining, and ontological building aides that overcome the challengeswill be discussed and analyzed.

Scientists always face difficulties dealing with disjointed information. There is a need for a standardized and robust way to represent and exchange knowledge. Ontology has been widely used for this purpose. However, since research involves semantics and operations, we need to conceptualize both of them. In this thesis, we propose ReShare to provide a solution for this problem. Maximizing utilization while preserving the semantics is one of the main challenges when the heterogeneous knowledge is combined. Therefore, operational annotations were designed to allow generic object modeling, binding and representation. Furthermore, a test bed is developed and preliminary results are presented to show the usefulness and robustness of our approach. Moreover, two aggregation techniques for fusing ontology matchers are investigated as an initial work for building an algorithm which converts descriptive ontologies into operational ones.

Les sciences de la vie produisent de grandes masses de données (par exemple, des essais cliniques et des articles scientifiques). L'intégration et l'analyse des différentes bases de données liées à la même question de recherche, par exemple la corrélation entre phénotypes et génotypes, sont essentielles pour découvrir de nouvelles connaissances. Pour cela, la communauté des sciences de la vie a adopté les techniques du Web sémantique pour réaliser l'intégration et l'interopérabilité des données, en particulier les ontologies. En effet, les ontologies représentent la brique de base pour représenter et partager la quantité croissante de données sur le Web. Elles fournissent un vocabulaire commun pour les humains, et des définitions d'entités formelles pour les machines.Un grand nombre d'ontologies et de terminologies biomédicales a été développé pour représenter et annoter les différentes bases de données existantes. Cependant, celles qui sont représentées avec différentes ontologies qui se chevauchent, c'est à dire qui ont des parties communes, ne sont pas interopérables. Il est donc crucial d'établir des correspondances entre les différentes ontologies utilisées, ce qui est un domaine de recherche actif connu sous le nom d'alignement d'ontologies.Les premières méthodes d'alignement d'ontologies exploitaient principalement le contenu lexical et structurel des ontologies à aligner. Ces méthodes sont moins efficaces lorsque les ontologies à aligner sont fortement hétérogènes lexicalement, c'est à dire lorsque des concepts équivalents sont décrits avec des labels différents. Pour pallier à ce problème, la communauté d'alignement d'ontologies s'est tournée vers l'utilisation de ressources de connaissance externes en tant que pont sémantique entre les ontologies à aligner. Cette approche soulève plusieurs nouvelles questions de recherche, notamment : (1) la sélection des ressources de connaissance à utiliser, (2) l'exploitation des ressources sélectionnées pour améliorer le résultat d'alignement. Plusieurs travaux de recherche ont traité ces problèmes conjointement ou séparément. Dans notre thèse, nous avons fait une revue systématique et une comparaison des méthodes proposées dans la littérature. Puis, nous nous sommes intéressés aux deux questions.Les ontologies, autres que celles à aligner, sont les ressources de connaissance externes (Background Knowledge : BK) les plus utilisées. Les travaux apparentés sélectionnent souvent un ensemble d'ontologies complètes en tant que BK même si, seuls des fragments des ontologies sélectionnées sont réellement efficaces pour découvrir de nouvelles correspondances. Nous proposons une nouvelle approche qui sélectionne et construit une ressource de connaissance à partir d'un ensemble d'ontologies. La ressource construite, d'une taille réduite, améliore, comme nous le démontrons, l'efficience et l'efficacité du processus d'alignement basé sur l'exploitation de BK.L'exploitation de BK dans l'alignement d'ontologies est une épée à double tranchant : bien qu'elle puisse augmenter le rappel (i.e., aider à trouver plus de correspondances correctes), elle peut réduire la précision (i.e., générer plus de correspondances incorrectes). Afin de faire face à ce problème, nous proposons deux méthodes pour sélectionner les correspondances les plus pertinentes parmi les candidates qui se basent sur : (1) un ensemble de règles et (2) l'apprentissage automatique supervisé. Nous avons expérimenté et évalué notre approche dans le domaine biomédical, grâce à la profusion de ressources de connaissances en biomédecine (ontologies, terminologies et alignements existants). Nous avons effectué des expériences intensives sur deux benchmarks de référence de la campagne d'évaluation de l'alignement d'ontologie (OAEI). Nos résultats confirment l'efficacité et l'efficience de notre approche et dépassent ou rivalisent avec les meilleurs résultats obtenus. / Life sciences produce a huge amount of data (e.g., clinical trials, scientific articles) so that integrating and analyzing all the datasets related to a given research question like the correlation between phenotypes and genotypes, is a key element for knowledge discovery. The life sciences community adopted Semantic Web technologies to achieve data integration and interoperability, especially ontologies which are the key technology to represent and share the increasing amount of data on the Web. Indeed, ontologies provide a common domain vocabulary for humans, and formal entity definitions for machines.A large number of biomedical ontologies and terminologies has been developed to represent and annotate various datasets. However, datasets represented with different overlapping ontologies are not interoperable. It is therefore crucial to establish correspondences between the ontologies used; an active area of research known as ontology matching.Original ontology matching methods usually exploit the lexical and structural content of the ontologies to align. These methods are less effective when the ontologies to align are lexically heterogeneous i.e., when equivalent concepts are described with different labels. To overcome this issue, the ontology matching community has turned to the use of external knowledge resources as a semantic bridge between the ontologies to align. This approach arises several new issues mainly: (1) the selection of these background resources, (2) the exploitation of the selected resources to enhance the matching results. Several works have dealt with these issues jointly or separately. In our thesis, we made a systematic review and historical evaluation comparison of state-of-the-art approaches.Ontologies, others than the ones to align, are the most used background knowledge resources. Related works often select a set of complete ontologies as background knowledge, even if, only fragments of the selected ontologies are actually effective for discovering new mappings. We propose a novel BK-based ontology matching approach that selects and builds a knowledge resource with just the right concepts chosen from a set of ontologies. The conducted experiments showed that our BK selection approach improves efficiency without loss of effectiveness.Exploiting background knowledge resources in ontology matching is a double-edged sword: while it may increase recall (i.e., retrieve more correct mappings), it may lower precision (i.e., produce more incorrect mappings). We propose two methods to select the most relevant mappings from the candidate ones: (1) based on a set of rules and (2) with Supervised Machine Learning. We experiment and evaluate our approach in the biomedical domain, thanks to the profusion of knowledge resources in biomedicine (ontologies, terminologies and existing alignments).We evaluated our approach with extensive experiments on two Ontology Alignment Evaluation Initiative (OAEI) benchmarks. Our results confirm the effectiveness and efficiency of our approach and overcome or compete with state-of-the-art matchers exploiting background knowledge resources.

Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / A descoberta de correspondÃncias semÃnticas entre esquemas à uma importante tarefa para diversos domÃnios de aplicaÃÃes, tais como integraÃÃo de dados, data warehouse e mashup de dados. Na maioria dos casos, as fontes de dados envolvidas sÃo heterogÃneas e dinÃmicas, dificultando ainda mais a realizaÃÃo dessa tarefa. Ontologias vÃm sendo utilizadas no intuito de definir vocabulÃrios comuns usados para descrever os elementos dos esquemas envolvidos em uma determinada aplicaÃÃo. O problema de matching entre ontologias, ou ontology matching, consiste na descoberta de correspondÃncias entre os termos dos vocabulÃrios (representados por ontologias) usados entre as diversas aplicaÃÃes. As soluÃÃes propostas na literatura, apesar de serem totalmente automÃticas possuem natureza heurÃstica, podendo produzir resultados nÃo-satisfatÃrios. O problema se intensifica quando se lida com grandes fontes de dados. O objetivo deste trabalho à propor um mÃtodo para geraÃÃo e refinamento incremental de correspondÃncias entre ontologias. A abordagem proposta faz uso de tÃcnicas de filtragem de ontologias, bem como do feedback do usuÃrio para dar suporte à geraÃÃo e ao refinamento dessas correspondÃncias. Para fins de validaÃÃo, uma ferramenta foi desenvolvida e experimentos foram realizados. / The discovery of semantic correspondences between schemas is an important task for different fields of applications such as data integration, data warehousing and data mashup. In most cases, the data sources involved are heterogeneous and dynamic, making it even harder the performance of that task. Ontologies are being used in order to define common vocabulary used to describe the elements of the schemas involved in a particular application. The problem of matching between ontologies, or ontology matching, consists in the discovery of correspondences between terms of vocabularies (represented by ontologies) used between the various applications. The solutions proposed in the literature, despite being fully automatic have heuristic nature, and may produce non-satisfactory results. The problem intensifies when dealing with large data sources. The purpose of this paper is to propose a method for generation and incremental refinement of correspondences between ontologies. The proposed approach makes use of filtering techniques of ontologies, as well as user feedback to support the generation and refining these matches. For validation purposes, a tool was developed and experiments were conducted

The World Wide Web contains large amounts of data, and in most cases this data has no explicit structure. The lack of structure makes it difficult for automated agents to understand and use such data. A step towards a more structured World Wide Web is the Semantic Web, which aims at introducing semantics to data on the World Wide Web. One of the key technologies in this endeavour are ontologies, which provide a means for modeling a domain of interest and are used for search and integration of data. In recent years many ontologies have been developed. To be able to use multiple ontologies it is necessary to align them, i.e., find inter-ontology relationships. However, developing and aligning ontologies is not an easy task and it is often the case that ontologies and their alignments are incorrect and incomplete. This can be a problem for semantically-enabled applications. Incorrect and incomplete ontologies and alignments directly influence the quality of the results of such applications, as wrong results can be returned and correct results can be missed. This thesis focuses on the problem of completing ontologies and ontology networks. The contributions of the thesis are threefold. First, we address the issue of completing the is-a structure and alignment in ontologies and ontology networks. We have formalized the problem of completing the is-a structure in ontologies as an abductive reasoning problem and developed algorithms as well as systems for dealing with the problem. With respect to the completion of alignments, we have studied system performance in the Ontology Alignment Evaluation Initiative, a yearly evaluation campaign for ontology alignment systems. We have also addressed the scalability of ontology matching, which is one of the current challenges, by developing an approach for reducing the search space when generating the alignment.Second, high quality completion requires user involvement. As users' time and effort are a limited resource we address the issue of limiting and facilitating user interaction in the completion process. We have conducted a broad study of state-of-the-art ontology alignment systems and identified different issues related to the process. We have also conducted experiments to assess the impact of user errors in the completion process. While the completion of ontologies and ontology networks can be done at any point in the life-cycle of ontologies and ontology networks, some of the issues can be addressed already in the development phase. The third contribution of the thesis addresses this by introducing ontology completion and ontology alignment into an existing ontology development methodology.

L'appariement des ontologies est une tâche primordiale pour palier au problème d'hétérogénéité sémantique et ainsi assurer une interopérabilité entre les applications utilisant différentes ontologies. Il consiste en la mise en correspondance de chaque entité d'une ontologie source à une entité d'une ontologie cible et ceci par application des techniques d'alignement fondées sur des mesures de similarité. Individuellement, aucune mesure de similarité ne permet d'obtenir un alignement parfait. C'est pour cette raison qu'il est intéressant de tenir compte de la complémentarité des mesures afin d'obtenir un meilleur alignement. Dans cette thèse, nous nous sommes intéressés à proposer un processus de décision crédibiliste pour l'appariement des ontologies. Étant données deux ontologies, on procède à leur appariement et ceci par application de trois techniques. Les alignements obtenus seront modélisés dans le cadre de la théorie des fonctions de croyance. Des règles de combinaison seront utilisées pour combiner les résultats d'alignement. Une étape de prise de décision s'avère utile, pour cette raison nous proposons une règle de décision fondée sur une distance et capable de décider sur une union d'hypothèses. Cette règle sera utilisée dans notre processus afin d'identifier pour chaque entité source le ou les entités cibles. / Ontology matching is a solution to mitigate the effect of semantic heterogeneity. Matching techniques, based on similarity measures, are used to find correspondences between ontologies. Using a unique similarity measure does not guarantee a perfect alignment. For that reason, it is necessary to use more than a similarity measure to take advantage of features of each one and then to combine the different outcomes. In this thesis, we propose a credibilistic decision process by using the theory of belief functions. First, we model the alignments, obtained after a matching process, under the theory of belief functions. Then, we combine the different outcomes through using adequate combination rules. Due to our awareness that making decision is a crucial step in any process and that most of the decision rules of the belief function theory are able to give results on a unique element, we propose a decision rule based on a distance measure able to make decision on union of elements (i.e. to identify for each source entity its corresponding target entities).

This dissertation describes WikiRank, an unsupervised method of assigning relative values to elements of a broad coverage encyclopedic information source in order to identify those entries that may be relevant to a given piece of text. The valuation given to an entry is based not on textual similarity but instead on the links that associate entries, and an estimation of the expected frequency of visitation that would be given to each entry based on those associations in context. This estimation of relative frequency of visitation is embodied in modifications to the random walk interpretation of the PageRank algorithm. WikiRank is an effective algorithm to support natural language processing applications. It is shown to exceed the performance of previous machine learning algorithms for the task of automatic topic identification, providing results comparable to that of human annotators. Second, WikiRank is found useful for the task of recognizing text-based paraphrases on a semantic level, by comparing the distribution of attention generated by two pieces of text using the encyclopedic resource as a common reference. Finally, WikiRank is shown to have the ability to use its base of encyclopedic knowledge to recognize terms from different ontologies as describing the same thing, and thus allowing for the automatic generation of mapping links between ontologies. The conclusion of this thesis is that the "knowledge access heuristic" is valuable and that a ranking process based on a large encyclopedic resource can form the basis for an extendable general purpose mechanism capable of identifying relevant concepts by association, which in turn can be effectively utilized for enumeration and comparison at a semantic level.

Ontology Matching is one of the hottest topic within the Semantic Web of recent years. There is still ample of space for improvement in terms of performance. Furthermore, current ontology matchers mostly concentrate on simple entity to entity matching. However, matching of whole structures could bring some additional complex relationships. These structures of ontologies can be captured as ontology patterns. The main theme of this thesis is an examination of pattern-based ontology matching enhanced with ontology transformation and pattern-based ontology alignment evaluation. The former is examined due to its potential benefits regarding complex matching and matching as such. The latter is examined because complex hypotheses could be beneficial feedback as complement to traditional evaluation methods. These two tasks are related to four different topics: ontology patterns, ontology transformation, ontology alignment evaluation and ontology matching. With regard to those four topics, this work covers the following aspects: * Examination of different aspects of ontology patterns. Particularly, description of relevant ontology patterns for ontology transformation and for ontology matching (such as naming, matching and transformation patterns). * Description of a pattern-based method for ontology transformation. * Introduction of new methods for an alignment evaluation; including using patterns as a complex structures for more detailed analysis. * Experiments and demonstrations of new concepts introduced in this thesis. The thesis first introduces naming pattern and matching pattern classification on which ontology transformation framework is based. Naming patterns are useful for detection of ontology patterns and for generation of new names for entities. Matching patterns are basis for transformation patterns in terms of sharing some building blocks. In comparison with matching patterns, transformation patterns have transformation links that represent way how parts of ontology patterns are transformed. Besides several evaluations and implementations, the thesis provides a demonstration of getting complex matching due to ontology transformation process. Ontology transformation framework has been implemented in Java environment where all generic patterns are represented as corresponding Java objects. Three main implemented services are made generally available as RESTful services: ontology pattern detection, transformation instruction generation and ontology transformation.

Ces dernières années, les ontologies ont suscité de nombreux travaux dans le domaine du web sémantique. Elles sont utilisées pour fournir le vocabulaire sémantique permettant de rendre la connaissance du domaine disponible pour l'échange et l'interprétation au travers des systèmes d'information. Toutefois, en raison de la nature décentralisée du web sémantique, les ontologies sont très hétérogènes. Cette hétérogénéité provoque le problème de la variation de sens ou ambiguïté dans l'interprétation des entités et, par conséquent, elle empêche le partage des connaissances du domaine. L'alignement d'ontologies, qui a pour but la découverte des correspondances sémantiques entre des ontologies, devient une tâche cruciale pour résoudre ce problème d'hétérogénéité dans les applications du web sémantique. Les principaux défis dans le domaine de l'alignement d'ontologies ont été décrits dans des études récentes. Parmi eux, la sélection de mesures de similarité appropriées ainsi que le réglage de la configuration de leur combinaison sont connus pour être des problèmes fondamentaux que la communauté doit traiter. En outre, la vérification de la cohérence sémantique des correspondances est connue pour être une tâche importante. Par ailleurs, la difficulté du problème augmente avec la taille des ontologies. Pour faire face à ces défis, nous proposons dans cette thèse une nouvelle approche, qui combine différentes techniques issues des domaines de l'apprentissage automatique, d'appariement de graphes et de recherche d'information en vue d'améliorer la qualité de l'alignement d'ontologies. En effet, nous utilisons des techniques de recherche d'information pour concevoir de nouvelles mesures de similarité efficaces afin de comparer les étiquettes et les profils d'entités de contexte au niveau des entités. Nous appliquons également une méthode d'appariement de graphes appelée propagation de similarité au niveau de la structure qui découvre effectivement des correspondances en exploitant des informations structurelles des entités. Pour combiner les mesures de similarité au niveau des entités, nous transformons la tâche de l'alignement d'ontologie en une tâche de classification de l'apprentissage automatique. Par ailleurs, nous proposons une méthode dynamique de la somme pondérée pour combiner automatiquement les correspondances obtenues au niveau des entités et celles obtenues au niveau de la structure. Afin d'écarter les correspondances incohérentes, nous avons conçu une nouvelle méthode de filtrage sémantique. Enfin, pour traiter le problème de l'alignement d'ontologies à large échelle, nous proposons deux méthodes de sélection des candidats pour réduire l'espace de calcul.Toutes ces contributions ont été mises en œuvre dans un prototype nommé YAM++. Pour évaluer notre approche, nous avons utilisé des données du banc d'essai de la compétition OAEI : Benchmark, Conference, Multifarm, Anatomy, Library and Large Biomedical Ontologies. Les résultats expérimentaux montrent que les méthodes proposées sont très efficaces. De plus, en comparaison avec les autres participants à la compétition OAEI, YAM++ a montré sa compétitivité et a acquis une position de haut rang. / In recent years, ontologies have attracted a lot of attention in the Computer Science community, especially in the Semantic Web field. They serve as explicit conceptual knowledge models and provide the semantic vocabularies that make domain knowledge available for exchange and interpretation among information systems. However, due to the decentralized nature of the semantic web, ontologies are highlyheterogeneous. This heterogeneity mainly causes the problem of variation in meaning or ambiguity in entity interpretation and, consequently, it prevents domain knowledge sharing. Therefore, ontology matching, which discovers correspondences between semantically related entities of ontologies, becomes a crucial task in semantic web applications.Several challenges to the field of ontology matching have been outlined in recent research. Among them, selection of the appropriate similarity measures as well as configuration tuning of their combination are known as fundamental issues that the community should deal with. In addition, verifying the semantic coherent of the discovered alignment is also known as a crucial task. Furthermore, the difficulty of the problem grows with the size of the ontologies. To deal with these challenges, in this thesis, we propose a novel matching approach, which combines different techniques coming from the fields of machine learning, graph matching and information retrieval in order to enhance the ontology matching quality. Indeed, we make use of information retrieval techniques to design new effective similarity measures for comparing labels and context profiles of entities at element level. We also apply a graph matching method named similarity propagation at structure level that effectively discovers mappings by exploring structural information of entities in the input ontologies. In terms of combination similarity measures at element level, we transform the ontology matching task into a classification task in machine learning. Besides, we propose a dynamic weighted sum method to automatically combine the matching results obtained from the element and structure level matchers. In order to remove inconsistent mappings, we design a new fast semantic filtering method. Finally, to deal with large scale ontology matching task, we propose two candidate selection methods to reduce computational space.All these contributions have been implemented in a prototype named YAM++. To evaluate our approach, we adopt various tracks namely Benchmark, Conference, Multifarm, Anatomy, Library and Large BiomedicalOntologies from the OAEI campaign. The experimental results show that the proposed matching methods work effectively. Moreover, in comparison to other participants in OAEI campaigns, YAM++ showed to be highly competitive and gained a high ranking position.

Submitted by Caroline Xavier (caroline.xavier@pucrs.br) on 2017-06-30T17:33:51Z No. of bitstreams: 1 TES_BERNARDO_SEVERO_DE_SOUZA_COMPLETO.pdf: 5113763 bytes, checksum: a24628d427a0a60b3a6ea0c5200d5dfd (MD5) / Made available in DSpace on 2017-06-30T17:33:51Z (GMT). No. of bitstreams: 1 TES_BERNARDO_SEVERO_DE_SOUZA_COMPLETO.pdf: 5113763 bytes, checksum: a24628d427a0a60b3a6ea0c5200d5dfd (MD5) Previous issue date: 2017-02-20 / O aumento do volume de dados n?o estruturados na Web nas ?ltimas d?cadas tem sido impulsionado pelo surgimento de novos meios de comunica??o, dispositivos e tecnologias. Neste contexto se desenvolve a Web Sem?ntica, cujo objetivo ? o de atribuir uma camada de representa??o de conhecimento a esses dados, facilitando o tratamento por processos automatizados. Ontologias s?o elementos chave da Web Sem?ntica, oferecendo uma descri??o dos conceitos e dos relacionamentos entre os mesmos para um dom?nio espec?fico. Entretanto, ontologias de um mesmo dom?nio podem divergir em sua estrutura, granularidade ou terminologia, necessitando que um processo de mapeamento entre as mesmas seja realizado, produzindo um conjunto de correspond?ncias entre entidades semanticamente relacionadas (alinhamento). Um n?mero crescente de abordagens de mapeamento tem surgido na literatura e a necessidade de avaliar e comparar qualitativamente os alinhamentos produzidos se faz presente. Tarefas que fazem uso de alinhamentos passaram a demandar melhores representa??es gr?ficas dos mesmos. Neste contexto, foi realizada uma pesquisa com especialistas em alinhamentos para identificar os aspectos mais importantes em uma visualiza??o de alinhamentos. Este trabalho apresenta ent?o uma abordagem adaptativa de visualiza??o para alinhamentos, que permite ao usu?rio escolher como e o que visualizar, de acordo com prefer?ncias pr?prias ou para uma atividade sendo realizada no momento (cria??o, manipula??o, avalia??o, etc.). Por fim, um prot?tipo foi constru?do com o intuito de validar a solu??o. Os resultados obtidos da avalia??o dos usu?rios com o prot?tipo mostram que a abordagem lida com os problemas que se prop?e a resolver, com uma margem para trabalhos futuros em formas de visualiza??o de alinhamentos. / The increase in the volume of unstructured web data in recent decades has been driven by the arising of new media, devices and technologies. In this context, the Semantic Web was developed, whose objective is to provide a layer of knowledge representation to that data, facilitating the treatment by automated processes. Ontologies are key elements of the Semantic Web, providing a description of the concepts and relationships between them, for a specific domain. However, ontologies of the same domain may differ in structure, granularity or terminology, requiring a process of matching between them to be performed, producing a set of correspondences between semantically related entities (alignment). A growing number of matching approaches have emerged in the literature, and the need to evaluate and qualitatively compare the produced alignments is presented. Tasks that make use of alignments started to demand better graphical representations for it. In this context, a survey was conducted with alignment specialists to identify the most important aspects in an alignment visualization. This work presents an adaptative approach for alignment visualization, that allows users to choose how and what to visualize, according to their own preferences or the task being performed at that moment (creation, manipulation, evaluation, etc.). Finally, a prototype was built with the purpose of validating the solution. The results obtained from the prototype validation with users show that the approach handles the problems it proposes to solve, with a margin for future work on alignment visualization.