Made available in DSpace on 2014-06-12T15:51:21Z (GMT). No. of bitstreams: 1 license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2008 / O reuso de software é uma área de pesquisa da engenharia de software que tem por objetivo prover melhorias na produtividade e qualidade da aplicação através da redução do esforço. Trata-se de reutilizar artefatos existentes, ao invés de construí-los do zero a fim de criar novas aplicações. Porém, para obter os benefícios inerentes ao reuso, alguns obstáculos devem ser superados como, por exemplo, a questão da busca e recuperação de componentes. Em geral, há uma lacuna entre a formulação do problema, na mente do desenvolvedor e a recuperação do mesmo no repositório, o que resulta em resultados irrelevantes diminuindo as chances de reuso. Dessa maneira, mecanismos que auxiliem na formulação das consultas e que contribuam para uma recuperação mais próxima à necessidade do desenvolvedor, são bastante oportunos para solucionar os problemas apresentados. Nesse contexto, este trabalho propõe a extensão de uma ferramenta de busca por palavra-chave através de uma camada semântica que tem por objetivo principal aumentar a precisão da busca e, conseqüentemente, aumentar as chances de reuso do componente procurado. A criação da camada semântica é representada basicamente por dois componentes principais: um para auxiliar o usuário na formulação da consulta, através do uso de uma ontologia de domínio, e outro para tornar a recuperação mais eficiente, através de uma indexação semântica dos componentes no repositório

The thesis investigates methods for uniform interpolation in expressive description logics. Description logics are formalisms commonly used to model ontologies. Ontologies store terminological information and are used in a wide range of applications, such as the semantic web, medicine, bio-informatics, software development, data bases and language processing. Uniform interpolation eliminates terms from an ontology such that logical entailments in the remaining language are preserved. The result, the uniform interpolant, is a restricted view of the ontology that can be used for a variety of tasks such as ontology analysis, ontology reuse, ontology evolution and information hiding. Uniform interpolation for description logics has only gained an interest in the research community in the last years, and theoretical results show that it is a hard problem requiring specialised reasoning approaches. We present a range of uniform interpolation methods that can deal with expressive description logics such as ALC and many of its extensions. For all these logics, these are the first methods that are able to compute uniform interpolants for all inputs. The methods are based a new family of saturation-based reasoning methods, which make it possible to eliminate symbols in a goal-oriented manner. The practicality of this approach is shown by an evaluation on realistic ontologies.

Biological knowledge is widely represented in the form of ontologies and ontology-based annotations. Biomedical ontologies describe known phenomena in biology using formal axioms, and the annotations associate an entity (e.g. genes, diseases, chemicals, etc.) with a set of biological concepts. In addition to formally structured axioms, ontologies contain meta-data in the form of annotation properties expressed mostly in natural language which provide valuable pieces of information that characterize ontology concepts. The structure and information contained in ontologies and their annotations make them valuable for use in machine learning, data analysis and knowledge extraction tasks. I develop the first approaches that can exploit all of the information encoded in ontologies, both formal and informal, to learn feature embeddings of biological concepts and biological entities based on their annotations to ontologies. Notably, I develop the first approach to use all the formal content of ontologies in the form of logical axioms and entity annotations to generate feature vectors of biological entities using neural language models. I extend the proposed algorithm by enriching the obtained feature vectors through representing the natural language annotation properties within the ontology meta-data as axioms. Transfer learning is then applied to learn from the biomedical literature and apply on the formal knowledge of ontologies. To optimize learning that combines the formal content of biomedical ontologies and natural language data such as the literature, I also propose a new approach that uses self-normalization with a deep Siamese neural network that improves learning from both the formal knowledge within ontologies and textual data. I validate the proposed algorithms by applying them to the Gene Ontology to generate feature vectors of proteins based on their functions, and to the PhenomeNet ontology to generate features of genes and diseases based on the phenotypes they are associated with. The generated features are then used to train a variety of machinelearning based classifiers to perform different prediction tasks including the prediction of protein interactions, gene–disease associations and the toxicological effects of chemicals. I also use the proposed methods to conduct the first quantitative evaluation of the quality of the axioms and meta-data included in ontologies to prove that including axioms as background improves ontology-based prediction. The proposed approaches can be applied to a wide range of other bioinformatics research problems including similarity-based prediction and classification of interaction types using supervised learning, or clustering.

Biological knowledge is widely represented in the form of ontologies and ontologybased annotations. Biomedical ontologies describe known phenomena in biology using formal axioms, and the annotations associate an entity (e.g. genes, diseases, chemicals, etc.) with a set of biological concepts. In addition to formally structured axioms, ontologies contain meta-data in the form of annotation properties expressed mostly in natural language which provide valuable pieces of information that characterize ontology concepts. The structure and information contained in ontologies and their annotations make them valuable for use in machine learning, data analysis and knowledge extraction tasks. I develop the rst approaches that can exploit all of the information encoded in ontologies, both formal and informal, to learn feature embeddings of biological concepts and biological entities based on their annotations to ontologies. Notably, I develop the rst approach to use all the formal content of ontologies in the form of logical axioms and entity annotations to generate feature vectors of biological entities using neural language models. I extend the proposed algorithm by enriching the obtained feature vectors through representing the natural language annotation properties within the ontology meta-data as axioms. Transfer learning is then applied to learn from the biomedical literature and apply on the formal knowledge of ontologies. To optimize learning that combines the formal content of biomedical ontologies and natural language data such as the literature, I also propose a new approach that uses self-normalization with a deep Siamese neural network that improves learning from both the formal knowledge within ontologies and textual data. I validate the proposed algorithms by applying them to the Gene Ontology to generate feature vectors of proteins based on their functions, and to the PhenomeNet ontology to generate features of genes and diseases based on the phenotypes they are associated with. The generated features are then used to train a variety of machinelearning based classi ers to perform di erent prediction tasks including the prediction of protein interactions, gene{disease associations and the toxicological e ects of chemicals. I also use the proposed methods to conduct the rst quantitative evaluation of the quality of the axioms and meta-data included in ontologies to prove that including axioms as background improves ontology-based prediction. The proposed approaches can be applied to a wide range of other bioinformatics research problems including similarity-based prediction and classi cation of interaction types using supervised learning, or clustering.

Submitted to the faculty of Indiana University in partial fulfillment of the requirements for the degree Masters of Science in the School of Informatics Indiana University December 2005 / This project addresses the need for an increased expressivity and robustness of ontologies already supporting BACIIS and SIBIOS, two systems for data and service integration in the life sciences. The previous ontology solutions as global schema and facilitator of service discovery sustained the purposes for which they were built to provide, but were in need of updating in order to keep up with more recent standards in ontology descriptions and utilization as well as increase the breadth of the domain and expressivity of the content. Thus, several tasks were undertaken to increase the worth of the system ontologies. These include an upgrade to a more recent ontology language standard, increased domain coverage, and increased expressivity via additions of relationships and hierarchies within the ontology as well as increased ease of maintenance by a distributed design.

Techniques for the industrial implementation of emerging semantic technologies are presented in this research. Every new design, project, and procedure within a company generates a considerable amount of new information and important knowledge. Furthermore, a tremendous amount of legacy knowledge already exists within companies in electronic and non-electronic formats. All of this generated knowledge results in the need for tools and techniques to represent, structure, and reuse this knowledge. Researchers have spent considerable time and effort developing semantic knowledge management systems, with anticipation that these tools will address these knowledge management needs. However, little has been done to implement these systems within an industrial setting. In this thesis, we identify five main requirements for the development of an industry-ready, semantic knowledge management system, and we discuss how each of these requirements can be methodically addressed. The five requirements include the incorporation of legacy information, the ease of new knowledge management software adoption, the robustness of the software to support multiple file types and allow for the sharing of information across platforms, the security of the stored information, and the ease of use of the user interface. In collaboration with Raytheon, a defense and aerospace systems company, we developed and demonstrated a novel approach for the successful adoption of semantic abilities by a commercial company. Salient features of this work include a new tool, the e-Design MemoExtractor Software Tool, custom designed to mine and capture company information, a Raytheon-specific ontology extension to the e-Design Framework, and a novel semantic environment in the form of a customized semantic media wiki SMW+. The advantages of this approach and the associated research issues are discussed in the context of the industrial case study with Raytheon.

No description available.

Seamless computer-based knowledge sharing between departments of a manufacturing enterprise is useful in preventing unnecessary design revisions. A lack of interoperability between independently developed knowledge bases, however, is a major impediment in the development of a seamless knowledge sharing system. Interoperability, being an ability to overcome semantic and syntactic differences during computer-based knowledge sharing can be enhanced through the use of ontologies. Ontologies in computer science terms are hierarchical structures of knowledge stored in a computer-based knowledge base. Ontologies have been accepted by all as an interoperable medium to provide a non-subjective way of storing and sharing knowledge across diverse domains. Some semantic and syntactic differences, however, still crop up when these ontological knowledge bases are developed independently. A case study in an aerospace components manufacturing company suggests that shape features of a component are perceived differently by the designing and manufacturing departments. These differences cause further misunderstanding and misinterpretation when computer-based knowledge sharing systems are used across the two domains. Foundation or core ontologies can be used to overcome these differences and to ensure a seamless sharing of knowledge. This is because these ontologies provide a common grounding for domain ontologies to be used by individual domains or department. This common grounding can be used by the mediation and knowledge verification systems to authenticate the meaning of knowledge understood across different domains. For this reason, this research proposes a knowledge verification framework for developing a system capable of verifying knowledge between those domain ontologies which are developed out of a common core or foundation ontology. This framework makes use of ontology logic to standardize the way concepts from a foundation and core-concepts ontology are used in domain ontologies and then by using the same principles the knowledge being shared is verified. The Knowledge Frame Language which is based on Common Logic is used for formalizing example ontologies. The ontology editor used for browsing and querying ontologies is the Integrated Ontology Development Environment (IODE) by Highfleet Inc. An ontological product modelling technique is also developed in this research, to test the proposed framework in the scenario of manufacturability analysis. The proposed framework is then validated through a Java API specially developed for this purpose. Real industrial examples experienced during the case study are used for validation.

De nombreux projets de numérisation sont actuellement menés en France et en Europe pour sauvegarder le contenu de dizaines de milliers de documents anciens. Les images de ces documents sont utilisées par les historiens pour identifier l’historique des livres. Cette thèse s’inscrit dans le projet Navidomass (ANR-06-MDCA-012) qui a pour but de valoriser le patrimoine écrit français de la Renaissance, en proposant d’identifier ses images pour les indexer. Dans le cadre de cette thèse, nous nous sommes particulièrement intéressés aux images graphiques. Ces documents,qui sont apparus avec le début de l’imprimerie, sont composées d’images complexes puisque composées de différentes couches d’informations (images de traits).Afin de répondre à ce problème, nous proposons un modèle ontologique d’analyse complexe d’images de documents anciens. Ce modèle permet d’intégrer dans une même base les connaissances propres aux historiens, et les connaissances extraites par des traitements d’images. De par la nature complexe de ces images, les méthodes habituelles d’analyse d’images et d’extraction automatique de connaissances sont inopérantes. Nous proposons donc une nouvelle approche d’analyse des images de documents anciens qui permet de les caractériser à partir de leurs spécificités. Cette approche commence par simplifier les images en les séparant en différentes couches d’informations (formes et traits). Puis, pour chaque couche, nous venons extraire des motifs utilisés pour décrire les images. Ainsi, les images sont caractérisées à l’aide de sacs de motifs fréquents, et de sacs de traits. Pour ces deux couches d’informations, nous venons également extraire des graphes de régions qui permettent d’extraire une connaissance structurelle des images. La complexification de ces deux descriptions est insérée dans la base de connaissances, pour permettre des requêtes complexes. Le but de cette base est de proposer à l’utilisateur de rechercher une image en indiquant soit un exemple d’images recherchées, soit des éléments caractéristiques des images. / In the general context of cultural heritage preservation campaigns, many digitization projects are being conducted in France and Europe to save the contents of thousands of ancient documents. Images of these documents are used by historians to identify the history of books. This thesis was led into the Navidomass project (ANR-06-MDCA-012) which aims at promoting the written heritage of the documents from the Renaissance, by proposing to identify its images. As part of this thesis, we are particularly interested in graphical images, and more specifically to dropcaps. These graphical images, which emerged with the beginning of printing, are complex images which can be seen as composed of different layers of information (images composed of strokes). To address this problem, we propose an ontological model of complex analysis of images of old documents. This model allows to integrate the knowledge specific to historians, and the knowledge extracted by image processing, into a single database. Due to the complex nature of these images, the usual methods of image analysis and automatic extraction of knowledge are inefficient. We therefore propose a new approach for analyzing images of old documents that can be characterized on their features basis. This approach begins by simplifying the images, separated in different layers of information (shapes and lines). Then, for each layer, we extract patterns used to describe the images. Thus, images are described with most common bags of patterns, and bags of stroke. For these two layers of information, we have also extracted graphs of regions that allow extracting a more structural knowledge of the images. A more complex description is then inserted into the knowledge base in order to allow complex queries. The purpose of this database is to offer the possiblity to make either query by example, or query by specific features of the images, to user.

Pour exploiter efficacement une masse toujours croissante de documents électroniques, une branche de l'Intelligence Artificielle s'est focalisée sur la création et l'utilisation de systèmes à base de connaissance. Ces approches ont prouvé leur efficacité, notamment en recherche d'information. Cependant elles imposent une indexation sémantique des ressources exploitées, i.e. que soit associé à chaque ressource un ensemble de termes qui caractérise son contenu. Pour s'affranchir de toute ambiguïté liée au langage naturel, ces termes peuvent être remplacés par des concepts issus d'une ontologie de domaine, on parle alors d'indexation conceptuelle.Le plus souvent cette indexation est réalisée en procédant à l'extraction des concepts du contenu même des documents. On note, dans ce cas, une forte dépendance des techniques associées à ce traitement au type de document et à l'utilisation d'algorithmes dédiés. Pourtant une des forces des approches conceptuelles réside dans leur généricité. En effet, par l'exploitation d'indexation sémantique, ces approches permettent de traiter de la même manière un ensemble d'images, de gènes, de textes ou de personnes, pour peu que ceux-ci aient été correctement indexés. Cette thèse explore ce paradigme de généricité en proposant des systèmes génériques et en les comparant aux approches existantes qui font référence. L'idée est de se reposer sur les annotations sémantiques et d'utiliser des mesures de similarité sémantique afin de créer des approches performantes. De telles approches génériques peuvent par la suite être enrichies par des modules plus spécifiques afin d'améliorer le résultat final. Deux axes de recherche sont suivis dans cette thèse. Le premier et le plus riche est celui de l'indexation sémantique. L'approche proposée exploite la définition et l'utilisation de documents proches en contenu pour annoter un document cible. Grâce à l'utilisation de similarités sémantiques entre les annotations des documents proches et à l'utilisation d'une heuristique, notre approche, USI (User-oriented Semantic Indexer), permet d'annoter des documents plus rapidement que les méthodes existantes en fournissant une qualité comparable. Ce processus a ensuite été étendu à une autre tâche, la classification. Le tri est une opération indispensable à laquelle l'Homme s'est attaché depuis l'Antiquité, qui est aujourd'hui de plus en plus automatisée. Nous proposons une approche de classification hiérarchique qui se base sur les annotations sémantiques des documents à classifier. Là encore, la méthode est indépendante des types de documents puisque l'approche repose uniquement sur leur annotations. Un autre avantage de cette approche est le fait que lorsque des documents sont rassemblés, le groupe qu'il forme est automatiquement annoté (suivant notre algorithme d'indexation). Par conséquent, le résultat fourni est une hiérarchie de classes contenant des documents, chaque classe étant annotée. Cela évite l'annotation manuelle fastidieuse des classes par l'exploration des documents qu'elle contient comme c'est souvent le cas.L'ensemble de nos travaux a montré que l'utilisation des ontologies permettait d'abstraire plusieurs processus et ainsi de réaliser des approches génériques. Cette généricité n'empêche en aucun cas d'être couplée à des approches plus spécifiques, mais constitue en soi une simplicité de mise en place dès lors que l'on dispose de documents annotés sémantiquement. / In order to improve the exploitation of even growing number of electronic documents, Artificial Intelligence has dedicated a lot of effort to the creation and use of systems grounded on knowledge bases. In particular in the information retrieval field, such semantic approaches have proved their efficiency.Therefore, indexing documents is a necessary task. It consists of associating them with sets of terms that describe their content. These terms can be keywords but also concepts from an ontology, in which case the annotation is said to be semantic and benefit from the inherent properties of ontologies which are the absence of ambiguities.Most approaches designed to annotate documents have to parse them and extract concepts from this parsing. This underlines the dependance of such approaches to the type of documents, since parsing requires dedicated algorithms.On the other hand, approaches that solely rely on semantic annotations can ignore the document type, enabling the creation of generic processes. This thesis capitalizes on genericity to build novel systems and compare them to state-of-the-art approaches. To this end, we rely on semantic annotations coupled with semantic similarity measures. Of course, such generic approaches can then be enriched with type-specific ones, which would further increase the quality of the results.First of all, this work explores the relevance of this paradigm for indexing documents. The idea is to rely on already annotated close documents to annotate a target document. We define a heuristic algorithm for this purpose that uses the semantic annotations of these close documents and semantic similarities to provide a generic indexing method. This results in USI (User-oriented Semantic Indexer) that we show to perform as well as best current systems while being faster.Second of all, this idea is extended to another task, clustering. Clustering is a very common and ancient process that is very useful for finding documents or understanding a set of documents. We propose a hierarchical clustering algorithm that reuses the same components of classical methods to provide a novel one applicable to any kind of documents. Another benefit of this approach is that when documents are grouped together, the group can be annotated by using our indexing algorithm. Therefore, the result is not only a hierarchy of clusters containing documents as clusters are actually described by concepts as well. This helps a lot to better understand the results of the clustering.This thesis shows that apart from enhancing classical approaches, building conceptual approaches allows us to abstract them and provide a generic framework. Yet, while bringing easy-to-set-up methods – as long as documents are semantically annotated –, genericity does not prevent us from mixing these methods with type-specific ones, in other words creating hybrid methods.