Doctor of Philosophy (PhD) / This thesis presents a computational method for the inductive inference of explicit and implicit semantic design knowledge from the symbolic-mathematical syntax of design formulations using an unsupervised pattern recognition and extraction approach. Existing research shows that AI / machine learning based design computation approaches either require high levels of knowledge engineering or large training databases to acquire problem reformulation knowledge. The method presented in this thesis addresses these methodological limitations. The thesis develops, tests, and evaluates ways in which the method may be employed for design problem reformulation. The method is based on the linear algebra based factorization method Singular Value Decomposition (SVD), dimensionality reduction and similarity measurement through unsupervised clustering. The method calculates linear approximations of the associative patterns of symbol cooccurrences in a design problem representation to infer induced coupling strengths between variables, constraints and system components. Unsupervised clustering of these approximations is used to identify useful reformulations. These two components of the method automate a range of reformulation tasks that have traditionally required different solution algorithms. Example reformulation tasks that it performs include selection of linked design variables, parameters and constraints, design decomposition, modularity and integrative systems analysis, heuristically aiding design “case” identification, topology modeling and layout planning. The relationship between the syntax of design representation and the encoded semantic meaning is an open design theory research question. Based on the results of the method, the thesis presents a set of theoretical postulates on computable relationships between design syntax and semantics. The postulates relate the performance of the method with empirical findings and theoretical insights provided by cognitive neuroscience and cognitive science on how the human mind engages in symbol processing and the resulting capacities inherent in symbolic representational systems to encode “meaning”. The performance of the method suggests that semantic “meaning” is a higher order, global phenomenon that lies distributed in the design representation in explicit and implicit ways. A one-to-one local mapping between a design symbol and its meaning, a largely prevalent approach adopted by many AI and learning algorithms, may not be sufficient to capture and represent this meaning. By changing the theoretical standpoint on how a “symbol” is defined in design representations, it was possible to use a simple set of mathematical ideas to perform unsupervised inductive inference of knowledge in a knowledge-lean and training-lean manner, for a knowledge domain that traditionally relies on “giving” the system complex design domain and task knowledge for performing the same set of tasks.

This thesis is linked to the research program developed at LABLER which stands for Laboratório de Pesquisa e Ensino de Leitura e Redação, a laboratory of research and reading and writing teaching and to the research group Language as social practice as it integrates the development of the CNPq Project (301962/2007- 3) Critical genre analysis with focus on science popularization articles (MOTTAROTH, 2007). As part of the project mentioned before, this thesis has as objective to examine the discursive practices involved in the midiatization of scientific knowledge from a linguistic and discursive point of view with attention to the way the content of a scientific article is resignified in science popularization news in a special way by the use of linguistic strategies of reformulation. The work is founded on the Critical Genre Analysis using as sources the Socio Rhetorical Perspective and the Critical Discourse Analysis and searches subsidies in the approach of the Pedagogical Model, by Bernstein (1996) to comprehend the science popularization process, according to Motta-Roth (2009a, b, c, d; 2010a, b) and Motta-Roth; Marcuzzo (2010) formulations. The research corpus is compounded by 60 science popularization news, published in the electronic media in the sites of BBC News International, Scientific American, Nature and ABC Science from 2004 to 2008. The contextual analysis of interpretative basis was made considering information gathered on the sites and their owners via hyperlinks. In the textual analysis, of interpretative basis, the reformulation cases were raised according to the adequacy of the taxonomy proposed by Hyland (2007) for the discursive functions of reformulation. The distribution and function of reformulation strategies in science popularization news had as basis the representation of the rhetorical organization of science popularization news genre elaborated by Motta-Roth and Lovato (2009, p. 246). The results point to the use of reformulation strategies as an essential aspect to recontextualization of scientific knowledge for allowing knowledge midiatization. This fulfillment is guaranteed by the strategies of meaning expansion and delimitation which allow that the writer guides the reader in the news reading process, characterizing it as an activity of mutual collaboration. / Esta tese está vinculada ao programa de pesquisa desenvolvido no LABLER Laboratório de Pesquisa e Ensino de Leitura e Redação da UFSM, e ao grupo de pesquisa Linguagem como prática social ao integrar o desenvolvimento do Projeto CNPq (nº 301962/2007-3) Análise crítica de gêneros com foco em artigos de popularização da ciência (MOTTA-ROTH, 2007). Como um recorte desse projeto, esta tese tem como objetivo examinar as práticas discursivas envolvidas na midiatização do conhecimento científico sob o viés linguístico-discursivo, com atenção ao modo como o conteúdo de um artigo científico é ressignificado em uma notícia de popularização da ciência (doravante PC), de modo especial pelo emprego de estratégias linguísticas de reformulação. O trabalho está fundamentado na Análise Crítica de Gênero complementada por subsídios da Sociorretórica e da Análise Crítica do Discurso e busca na abordagem do modelo de discurso pedagógico de Bernstein (1996), subsídios para a compreensão do processo de PC, conforme as formulações de Motta-Roth (2009a, b, c, d; 2010a, b) e Motta-Roth; Marcuzzo (2010). O corpus da pesquisa é composto por 60 notícias de PC, publicadas na mídia eletrônica, nos sites da BBC News International, Scientific American, Nature e ABC Science no período de 2004 a 2008. A análise contextual, de base interpretativa, foi feita a partir das informações disponíveis nos sites das revistas on-line e de seus proprietários, indicados via hiperlinks. Na análise textual, de base interpretativa, os casos de reformulação foram levantados a partir da adequação da abordagem da representação das funções discursivas de reformulação, proposta por HYLAND (2007). A distribuição e a função das estratégias de reformulação nas notícias de PC teve como base a representação esquemática do gênero notícia de PC elaborada por Motta-Roth e Lovato (2009, p. 246). Os resultados apontam o emprego da reformulação como aspecto essencial à recontextualização do conhecimento cientifico ao permitir a efetivação do movimento de midiatização do conhecimento. Essa efetivação é garantida pelas estratégias de expansão e delimitação do sentido que permitem que o escritor oriente o leitor no processo de leitura da notícia de PC, caracterizando-a como uma atividade de colaboração mútua.

Submitted by Daniella Sodre (daniella.sodre@ufpe.br) on 2015-03-10T13:03:19Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação Bruno Souza.pdf: 3055649 bytes, checksum: 5cedaf83e4e87135a1f22f1bb7c1dd09 (MD5) / Made available in DSpace on 2015-03-10T13:03:19Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação Bruno Souza.pdf: 3055649 bytes, checksum: 5cedaf83e4e87135a1f22f1bb7c1dd09 (MD5) Previous issue date: 2013-09-09 / FACEPE / Ambientes dinâmicos e distribuídos são sistemas descentralizados que fornecem aos usuários recursos de consultas sobre um conjunto de fontes de dados heterogêneas, distribuídas e autônomas (peers). Sistemas de Integração de Dados, Peer Data Management System (PDMS) e Dataspaces são exemplos de tais sistemas. Eles são constituídos por peers que pertencem a um domínio específico e estão ligados entre si por meio de correspondências semânticas. No entanto, um desafio inerente em ambientes dinâmicos e distribuídos é o processo de reformulação de consulta entre um par de peers. Quando um usuário coloca uma consulta em um peer, a fim de adquirir mais informações, a consulta deve ser reformulada de acordo com o esquema dos peers vizinhos. Neste processo podem surgir alguns problemas como a perda semântica e a degradação da consulta. A perda semântica e degradação da consulta são problemas relacionados à perda de conceitos semânticos durante a reformulação. Por outro lado, em um ambiente semanticamente rico, ao invés de uma perda semântica, a consulta pode ter um enriquecimento semântico por meio da agregação de conceitos semanticamente relacionados durante a reformulação. Neste sentido, a consulta do usuário pode ser enriquecida e resultados semânticos mais ricos podem ser recuperados. Critérios de qualidade da informação têm sido usados em alguns trabalhos para avaliar o nível de qualidade dos elementos de um ambiente dinâmico e distribuído como, por exemplo, peers, dados e a resposta da consulta. Estes critérios são medidas dinâmicas proporcionadas pelo sistema e servem como uma pontuação que pode ser constantemente avaliada para obter o nível real de qualidade. Neste trabalho, apresentamos quatro critérios de qualidade da informação que medem a perda e o ganho de conceitos semânticos durante a reformulação da consulta entre os pares de peers. Nós apresentamos um exemplo da nossa abordagem e os algoritmos de avaliação de critérios. Também damos as nossas definições para os problemas de perda semântica e degradação da consulta. Por fim, apresentamos a experimentação que fizemos com o PDMS SPEED e os resultados obtidos. / Dynamic distributed environments are decentralized systems that provide users with querying capabilities over a set of heterogeneous, distributed and autonomous data sources (peers). Data Integration Systems, Peer Data Management Systems (PDMS) and Dataspaces are examples of such systems. They are composed by peers that belong to a specific domain and are linked to each other by correspondences (semantic connections). Nonetheless, a challenge inherent to dynamic distributed environments is the query reformulation process between a pair of peers. When a user poses a query at a peer, in order to acquire more information, the query should be reformulated in accordance with the neighbor peers schema. In this process some problems as semantic loss and query degradation can arise. The semantic loss and query degradation are problems related to the loss of semantic concepts during query reformulation. In the other hand, in such a semantic environment instead of a semantic loss the query can have a semantic enrichment by aggregating semantic related concepts during reformulation. In this sense, the user’s query can be enriched and semantically richer results can be delivered. Information Quality criteria has been used in some works to evaluate the level of quality of the distributed dynamic environment’s elements such as, peers, data and query answer. These criteria are dynamic measures provided by the system and serve as scores that can be constantly evaluated to get the actual level of quality. In this work we present four Information Quality criteria that measure the loss and enrichment of semantic concepts during query reformulation among peers. We present an example of our analysis and the algorithms that implement the evaluation of the presented criteria. We also give our definitions to the semantic loss and query degradation problems. Finally, we present the experimentation we have done with the SPEED PDMS and the obtained results.

A dinâmica do mercado e a constante reformulação de produtos alimentícios são contínuas, porém a monitoração das alterações na composição química é escassa. Ao mesmo tempo, as bases de dados de composição de alimentos devem ser continuamente atualizadas. Os objetivos desse trabalho foram realizar a avaliação comparativa do perfil nutricional de grupos de alimentos específicos no período de 2003 e 2013, bem como a sistematização do processo de decisão de produtos prioritários para atualização de bases de dados de composição química de alimentos. Para a realização dos objetivos propostos foi necessário atualizar dados da Tabela Brasileira de Composição de Alimentos (TBCA) e compilar novas informações. O conteúdo de carboidratos, lipídios, proteínas, fibra alimentar (FA) e energia de 259 produtos pertencentes e quatro grupos de alimentos pré-estabelecidos foram estudados para a avaliação comparativa dos produtos com dados de 2003 e 2013. As técnicas de estatísticas multivariadas (agrupamento e componentes principais), variação percentual e a diferença percentual (D%) foram utilizadas. Através da análise de subgrupamento (cluster) os produtos foram divididos por similaridade na composição química e pela análise de componentes principais (ACP) foi possível verificar que o subagrupamento para os cereais e carnes ocorreu, principalmente, pelo conteúdo de proteínas e carboidratos, enquanto que para leites e produtos manufaturados foi devido aos carboidratos e lipídios. Na maioria dos subgrupos foi possível observar diferença significante em pelo menos um componente através do teste t-Student pareado. Essas alterações são compatíveis com o relevante número de produtos que apresentaram redução pela variação percentual de lipídios nos leites (53 %), bem como aumento de FA e lipídios nos cereais (55 %) e carnes (40 %), respectivamente. Para a decisão de alimentos prioritários para atualização foram desenvolvidas etapas para a sistematização. A primeira consistiu em avaliar a adequação da composição química dos produtos através do cálculo da D%, onde os alimentos foram classificados como \"dados compatíveis\" ou \"dados não compatíveis\", gerando a categorização dos produtos em diferentes graus de prioridades de atualização (muito elevada, elevada, média e baixa). Posteriormente, dentro dessas categorias, foi considerada a importância do alimento em relação a sua aquisição, com informações dos produtos e marcas mais compradas pela população. De 330 produtos categorizados 82 são prioritários para atualização. Portanto, as técnicas de estatística empregadas e a variação percentual permitiram avaliar a mudança no conteúdo de componentes tanto de forma individual quanto conjunta com demais nutrientes para os grupos de alimentos selecionados, além de reforçar a necessidade de periódica monitoração no perfil nutricional dos alimentos. Paralelamente, a sistematização criada auxiliou na decisão de alimentos prioritários para atualização, resultando em relevante redução do número de produtos para serem atualizados, sendo um modelo útil para bases de dados. / The market dynamics and the constant reformulation of food products are continuous. At the same time, the food composition databases should be continuously updated. The aims of this work was to perform a comparative evaluation of the nutritional profile of specific food groups in the period from 2003 to 2013, as well as to make the systematization of the decision-making process for priority food products to update the food composition databases. To achieve the proposed aims it was necessary update data from the Brazilian Food Composition Database (TBCA) and compile new information. The content of carbohydrates, lipids, proteins, dietary fiber (DF) and energy of 259 products distributed in four groups was studied for comparative evaluation of food products with data from 2003 and 2013. Multivariate statistical techniques (Cluster and Principal Component Analysis), percentage change and calculating of the percentage difference (D%) were used. Through the Cluster Analysis the food products were divided by similarities in food composition and by Principal Component Analysis (PCA) was evaluated that clusters occurred for cereals and meats groups, mostly, according to the proteins and carbohydrates content, and milks and manufactured foods groups according to carbohydrates and lipids content. In most clusters it was a significant difference was observed in at least in one component accord to paired t- Student test. These changes are compatible with the analysis of percentage change, the relevant number of products that presented a reduction in lipids in milks (53 %), as well as increase in DF and lipids in cereals (55 %) and meats (40 %), respectively. For the decision of the priority food products to update steps were developed for the systematization. The first consists in evaluating the chemical composition of the products through D%, where the products were classified as \"consistent data\" or \"non-consistent data\", thus the food products were categorized with different grades of priority (Very High, High, Medium and Low). After that, between these categories, the food products were evaluated for their importance in relation to their purchase, with product information and most-bought brands. From 330 food 82 were priority to update. Therefore, the joint techniques applied allowed nutrient content change to be assessed both in an individually-based manner as well as in a group for the selected food groups and support the need for periodic monitoring of the nutritional profile of foods. At the same time, the systematization created helped in decision-making of priority food products to update, which resulted in relevant decrease in the number of priority foods to be updated, being a useful model for food composition databases.

Cette thèse a pour objectif de donner une représentation du fonctionnement de la communication dans les discussions à visée philosophique (enregistrées à l’école primaire et au collège). Plus particulièrement, il s’agit d’étudier la dimension intersubjective de la communication et de la construction du sens au sein de ces pratiques ainsi que la façon dont l’interaction se co-construit. Selon nous, la reconnaissance de l’importance de l’interlocuteur dans l’énonciation est indispensable pour saisir la véritable nature de la communication. Ainsi, en partant d’une approche interactionniste et socio-constructiviste, la construction du sens est envisagée comme un processus dynamique et comme une action commune aux locuteur et interlocuteur(s). Les bénéfices des pratiques philosophiques avec les enfants sont reconnus. L’analyse des interactions menée sur corpus montre que la discussion à visée philosophique est un terrain propice à la construction collective du discours, du sens et de la conceptualisation. Nous nous intéressons aux mécanismes de cette construction collective réalisée par les enfants et au rôle de l’animateur dans cette construction. Nous étudions particulièrement les phénomènes de reprise et un phénomène linguistique figurant parmi les processus d’orientation de l’attention permettant de guider l’interprétation et la compréhension des interlocuteurs : la dislocation à gauche du sujet. L’observation des données permet de contribuer à définir les fonctions pragmatiques de ces phénomènes ainsi que les différents buts communicatifs associés à leur emploi par les enfants et par l’animateur lors des discussions à visée philosophique / This thesis investigates meaning construction during philosophical discussions that took place at both a French first and middle school. More precisely, it studies the intersubjective dimension of communication and the construction of meaning in these practices. At the same time, it examines the ways in which the interaction is co-constructed. We believe that recognising the interlocutor’s importance in the utterance act is essential in order to understand the true nature of the communication. Employing interactionist and socio-constructivist approaches, the construction of meaning is considered as a dynamic process and as a joint action between the speaker and interlocutor(s).The benefits of conducting philosophical discussions of this type with children have previously been recognised in the literature. The analysis of interactional corpora shows that philosophical discussion is a favourable environment for the collective construction of speech, meaning and conceptualization. This doctoral study considers the mechanisms of this collective construction conducted by children and the role of the facilitator within this process. In this way, the thesis examines different phenomena like the repetition/reformulation and the left-dislocation of the subject. This linguistic phenomenon forms part of the attention orientation process and guides collective interpretation and understanding. Data analysis allows the thesis to contribute to define the pragmatic functions of the studied phenomena and the different communication purposes associated with their use by children and by the facilitator during philosophical discussions.

Este trabalho realizado na Área de Estudos Linguísticos, Literários e Tradutológicos de Francês do Departamento de Letras Modernas da FFLCH/USP se situa na intersecção do Direito e das Ciências da Linguagem, numa perspectiva multi e interdisciplinar, do discurso jurídico. O corpus, formado por textos jurídicos em língua francesa e portuguesa (brasileira), tem como objeto de estudo o discurso jurídico. Nele conceitos de direito foram, ao longo do tempo, modificados e integrados às práticas sociais. O método comparativo foi o escolhido para melhor examinar as transformações. No primeiro texto, escolhido pela sua importância histórica, a Declaração de Direitos do Homem e do Cidadão (França, 1789), surgem os conceitos de homem, igualdade e liberdade. Esses mesmos conceitos reaparecem na Declaração Universal dos Direitos do Homem (ONU, 1948), em que direitos mais explicitados conferem maior proteção ao homem. O terceiro documento, a Constituição da República Federativa do Brasil (Brasil, 1988) detalha no seu artigo 5° todos os direitos conferidos ao homem do século XX, conceitos que ressurgem implícitos na lei sobre a Informatização dos Processos Judiciais (Brasil, lei 11.419/2006). A partir da observação desses documentos levanta-se a hipótese de que os conceitos apresentados de homem, liberdade, igualdade e de justiça não são os mesmos presentes no primeiro documento, o que evidencia uma evolução sócio-histórico-linguístico-discursiva. A reformulação ocasiona, inclusive, um apagamento de conceito. Dessa forma o liame que une os documentos é resultante de um processo. As noções de polifonia e de dialogismo emprestados da Análise do Discurso são desenvolvidos a partir dos estudos bakhtinianos e dos procedimentos argumentativos da retórica renovada por Perelman, bem como das reformulações discursivas implicadas. 8 No primeiro capítulo os documentos são contextualizados historicamente, a fim de que se possa entender seu surgimento. O segundo, teórico, verifica a presença da Teoria da Enunciação e dos conceitos de auditório, ethos e logos, integrantes da Teoria da Argumentação. Do ponto de vista linguístico-discursivo o que se valoriza são as reformulações como ferramentas dessas alterações. Ressalta-se que tais transformações tecem-se a partir das relações entre a Língua e o Direito. Os conceitos de transplant, circulation juridique e empréstimo propostos pelo Direito Comparado consideram as diferenças entre as línguas e as culturas francesa e brasileira. O terceiro capítulo trata do discurso jurídico e de uma proposta de classificação dos diferentes textos em subgêneros a partir de seu objetivo e local de produção. O resultado das análises revela que o decurso do tempo provoca a evolução dos conceitos de homem, liberdade, igualdade e justiça, a ponto de sua existência material no texto escrito permitir seu apagamento, passando de concreta à abstrata, já que implicitamente é o conteúdo que justifica a proposição e a publicação da lei Informatização dos Processos Judiciais. / This following thesis, developed in the French Linguistic, Literary and Translation area of the Modern Language Department FFLCH / USP lies in the intersection of Law and Language Sciences. The corpus composed of legal texts in French and Portuguese (Brazilian) has as its object of study of the legal discourse in which the concepts of law, over time, have been modified and integrated with social practices. The comparative method was chosen to better examine these transformations. The first text, chosen for its historical significance, the Declaration of Rights of Man and of the Citizen (1789, France), bring to the light the concepts of \"man\", \"equality\" and \"freedom\". These same concepts reappear in the Universal Declaration of Human Rights (UN, 1948), in which most rights specified give greater protection to the man. The third document, the Constitution of the Federative Republic of Brazil (Brazil, 1988) explains, in its 5th article, all rights granted to man of the 20th century, concepts which reappear implicit in the Law on the Informatized System of the Judicial Process (Brazil, law 11,419/2006). The observation of these documents raises the hypothesis that the concepts presented \"man\", \"freedom\", \"equality\" and \"justice\" are not the same in the first document, which highlights a socio-historical-linguistic evolution-discursive. The reformulation causes, also, an erase of concept. In this way the link that connects the documents is the result of a process. The concepts of polyphony and dialogism borrowed from Discourse Analysis are developed from Bakthins studies and procedures of argumentative renewed Rhetoric by Perelman, as well as discursives involved. In the first chapter documents are historically contextualized in order to permit their understanding and emergence. In the second chapter, theoretical, the Enunciation Theory and the concepts of dialogism and polyphony are noted in texts as well as the concepts of the Auditorium, ethos and logos from the 10 Theory of Argumentation. In the aspect of Linguistically-discursive cases the reformulations are the tools of these changes. It should be noted that such transformations weave from relations between the Languag and the Law. The concepts of \"transplant\", \"circulation juridique\" and \"borrowing\" proposed by the Comparative Law consider the differences between the French and Brazilian languages and cultures. The third chapter treat the legal discourse and propose a classification for texts in different sub-genres from its goal and production site. The result of the analysis prove that the course of time causes the evolution of the concepts of \"man\", \"freedom\", \"equality\" and \"justice\", to the point that it may even result in the erasure of his material existence in the written text, ranging from concret to abstract, since its content appears implicitly in the proposition and the publication of Law on the Informatized System of the Judicial Process.

This research is concerned with the development of algorithmic approaches for solving problems that arise in the design and analysis of telecommunication networks, location-allocation distribution contexts, and intelligent transportation networks. Specifically, the corresponding problems addressed in these areas are a local access and transport area (LATA) network design problem, the discrete equal-capacity p-median problem (PMED), and the estimation of dynamic origin-destination path ows or trip tables in a general network. For the LATA network problem, we develop a model and apply the Reformulation-Linearization Technique (RLT) to construct various enhanced tightened versions of the proposed model. We also design efficient Lagrangian dual schemes for solving the linear programming relaxation of the various enhanced models, and construct an effective heuristic procedure for deriving good quality solutions in this process. Extensive computational results are provided to demonstrate the progressive tightness resulting from the enhanced formulations and their effect on providing good quality feasible solutions. The results indicate that the proposed procedures typically yield solutions having an optimality gap of less than 2% with respect to the derived lower bound, within a reasonable effort that involves the solution of a single linear program. For the discrete equal-capacity p-median problem, we develop various valid inequalities, a separation routine for generating cutting planes via specific members of such inequalities, as well as an enhanced reformulation that constructs a partial convex hull representation that subsumes an entire class of valid inequalities via its linear programming relaxation. We also propose suitable heuristic schemes for solving this problem, based on sequentially rounding the continuous relaxation solutions obtained for the various equivalent formulations of the problem. Extensive computational results are provided to demonstrate the effectiveness of the proposed valid inequalities, enhanced formulations, and heuristic schemes. The results indicate that the proposed schemes for tightening the underlying relaxations play a significant role in enhancing the performance of both exact and heuristic solution methods for solving this class of problems. For the estimation of dynamic path ows in a general network, we propose a parametric optimization approach to estimate time-dependent path ows, or origin-destination trip tables, using available data on link traffic volumes for a general road network. Our model assumes knowledge of certain time-dependent link ow contribution factors that are a dynamic generalization of the path-link incidence matrix for the static case. We propose a column generation approach that uses a sequence of dynamic shortest path subproblems in order to solve this problem. Computational results are presented on several variants of two sample test networks from the literature. These results indicate the viability of the proposed approach for use in an on-line mode in practice. Finally, we present a summary of our developments and results, and offer several related recommendations for future research. / Ph. D.

Semantic graphs have been used for optimizing various natural language processing tasks as well as augmenting search and information retrieval tasks. In most cases these semantic graphs have been constructed through supervised machine learning methodologies that depend on manually curated ontologies such as Wikipedia or similar. In this thesis, which consists of two parts, we explore in the first part the possibility to automatically populate a semantic graph from an ad hoc data set of 50 000 newspaper articles in a completely unsupervised manner. The utility of the visual representation of the resulting graph is tested on 14 human subjects performing basic information retrieval tasks on a subset of the articles. Our study shows that, for entity finding and document similarity our feature engineering is viable and the visual map produced by our artifact is visually useful. In the second part, we explore the possibility to identify entity relationships in an unsupervised fashion by employing abstractive deep learning methods for sentence reformulation. The reformulated sentence structures are qualitatively assessed with respect to grammatical correctness and meaningfulness as perceived by 14 test subjects. We negatively evaluate the outcomes of this second part as they have not been good enough to acquire any definitive conclusion but have instead opened new doors to explore. / Semantiska grafer har använts för att optimera olika processer för naturlig språkbehandling samt för att förbättra sökoch informationsinhämtningsuppgifter. I de flesta fall har sådana semantiska grafer konstruerats genom övervakade maskininlärningsmetoder som förutsätter manuellt kurerade ontologier såsom Wikipedia eller liknande. I denna uppsats, som består av två delar, undersöker vi i första delen möjligheten att automatiskt generera en semantisk graf från ett ad hoc dataset bestående av 50 000 tidningsartiklar på ett helt oövervakat sätt. Användbarheten hos den visuella representationen av den resulterande grafen testas på 14 försökspersoner som utför grundläggande informationshämtningsuppgifter på en delmängd av artiklarna. Vår studie visar att vår funktionalitet är lönsam för att hitta och dokumentera likhet med varandra, och den visuella kartan som produceras av vår artefakt är visuellt användbar. I den andra delen utforskar vi möjligheten att identifiera entitetsrelationer på ett oövervakat sätt genom att använda abstraktiva djupa inlärningsmetoder för meningsomformulering. De omformulerade meningarna utvärderas kvalitativt med avseende på grammatisk korrekthet och meningsfullhet såsom detta uppfattas av 14 testpersoner. Vi utvärderar negativt resultaten av denna andra del, eftersom de inte har varit tillräckligt bra för att få någon definitiv slutsats, men har istället öppnat nya dörrar för att utforska.

Projection methods for the incompressible Navier-Stokes equations (NSE) are efficient, but introduce numerical boundary layers and have limited temporal accuracy due to their fractional step nature. The Pressure Poisson Equation (PPE) reformulations represent a class of methods that replace the incompressibility constraint by a Poisson equation for the pressure, with a suitable choice of the boundary condition so that the incompressibility is maintained. PPE reformulations of the NSE have important advantages: the pressure is no longer implicitly coupled to the velocity, thus can be directly recovered by solving a Poisson equation, and no numerical boundary layers are generated; arbitrary order time-stepping schemes can be used to achieve high order accuracy in time. In this thesis, we focus on numerical approaches of the PPE reformulations, in particular, the Shirokoff-Rosales (SR) PPE reformulation. Interestingly, the electric boundary conditions, i.e., the tangential and divergence boundary conditions, provided for the velocity in the SR PPE reformulation render classical nodal finite elements non-convergent. We propose two alternative methodologies, mixed finite element methods and meshfree finite differences, and demonstrate that these approaches allow for arbitrary order of accuracy both in space and in time. / Mathematics

L’objectif principal de la présente étude était de dresser un portrait développemental des reformulations orales d’apprenants adultes anglophones (n=124) ayant participé à un programme d’immersion à l’étranger (PIE) de cinq semaines en français langue seconde (FLS). Plus particulièrement, cette étude s’est servi des autoreformulations autoamorcées (ARAA) produites, c’est-à-dire les réparations du discours initiées personnellement par le locuteur et ce, lors de deux épreuves de narration (T1 et T2) pour comprendre les comportements d’ARAA en FLS chez ces apprenants, plus spécifiquement leur nature et leur précision. L’étude visait deux groupes de participants distincts, soit ceux qui étaient les plus débutants et les plus avancés en FLS lors de leur entrée au PIE, et cherchait à comprendre leur développement par rapport à la nature, la compétence et la précision de leurs ARAA. Les résultats ont montré une augmentation du nombre et de la précision des ARAA peu importe le niveau des apprenants.