Orientadores: Ricardo da Silva Torres, Anderson Rocha / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-24T22:15:52Z (GMT). No. of bitstreams: 1 Faria_FabioAugusto_D.pdf: 5657546 bytes, checksum: 5b95fa0f8a5653e7b13d8895cde208f1 (MD5) Previous issue date: 2014 / Resumo: O crescente aumento de dados visuais, seja pelo uso de inúmeras câmeras de vídeo monitoramento disponíveis ou pela popularização de dispositivos móveis que permitem pessoas criar, editar e compartilhar suas próprias imagens/vídeos, tem contribuído enormemente para a chamada ''big data revolution". Esta grande quantidade de dados visuais dá origem a uma caixa de Pandora de novos problemas de classificação visuais nunca antes imaginados. Tarefas de classificação de imagens e vídeos foram inseridos em diferentes e complexas aplicações e o uso de soluções baseadas em aprendizagem de máquina tornou-se mais popular para diversas aplicações. Entretanto, por outro lado, não existe uma ''bala de prata" que resolva todos os problemas, ou seja, não é possível caracterizar todas as imagens de diferentes domínios com o mesmo método de descrição e nem utilizar o mesmo método de aprendizagem para alcançar bons resultados em qualquer tipo de aplicação. Nesta tese, propomos um arcabouço para seleção e fusão de classificadores. Nosso método busca combinar métodos de caracterização de imagem e aprendizagem por meio de uma abordagem meta-aprendizagem que avalia quais métodos contribuem melhor para solução de um determinado problema. O arcabouço utiliza três diferentes estratégias de seleção de classificadores para apontar o menos correlacionados e eficazes, por meio de análises de medidas de diversidade. Os experimentos mostram que as abordagens propostas produzem resultados comparáveis aos famosos métodos da literatura para diferentes aplicações, utilizando menos classificadores e não sofrendo com problemas que afetam outras técnicas como a maldição da dimensionalidade e normalização. Além disso, a nossa abordagem é capaz de alcançar resultados eficazes de classificação usando conjuntos de treinamento muito reduzidos / Abstract: The frequent growth of visual data, either by countless available monitoring video cameras or the popularization of mobile devices that allow each person to create, edit, and share their own images and videos have contributed enormously to the so called ''big-data revolution''. This shear amount of visual data gives rise to a Pandora box of new visual classification problems never imagined before. Image and video classification tasks have been inserted in different and complex applications and the use of machine learning-based solutions has become the most popular approach to several applications. Notwithstanding, there is no silver bullet that solves all the problems, i.e., it is not possible to characterize all images of different domains with the same description method nor is it possible to use the same learning method to achieve good results in any kind of application. In this thesis, we aim at proposing a framework for classifier selection and fusion. Our method seeks to combine image characterization and learning methods by means of a meta-learning approach responsible for assessing which methods contribute more towards the solution of a given problem. The framework uses three different strategies of classifier selection which pinpoints the less correlated, yet effective, classifiers through a series of diversity measure analysis. The experiments show that the proposed approaches yield comparable results to well-known algorithms from the literature on many different applications but using less learning and description methods as well as not incurring in the curse of dimensionality and normalization problems common to some fusion techniques. Furthermore, our approach is able to achieve effective classification results using very reduced training sets / Doutorado / Ciência da Computação / Doutor em Ciência da Computação

Orientadores: Anna Christina Bentes da Silva, Edwiges Maria Morato / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Estudos da Linguagem / Made available in DSpace on 2018-08-26T03:57:47Z (GMT). No. of bitstreams: 1 Lima_RafahelJeanParintins_M.pdf: 2300492 bytes, checksum: 80c89200029eb9b405958ab9a7b3ec3e (MD5) Previous issue date: 2014 / Resumo: Nesta dissertação, discutimos o papel de processos cognitivo-discursivos na identificação de perspectivas sociais presentes em um grupo social composto por pesquisadores e afásicos. Objetivamos analisar movimentos convergentes e divergentes de ações cognitivo-discursivas realizadas em entrevistas de caráter temático (ALBERTI, 2004) com participantes do Centro de Convivência de Afásicos (CCA), situado no Instituto de Estudos da Linguagem (IEL/UNICAMP): um centro de interação social entre pessoas afásicas e não afásicas. Elaboramos um protocolo de entrevista que tematizou a definição, a organização grupal e a experiência individual e coletiva do CCA. Aplicamos esse protocolo em 2 (dois) sujeitos afásicos e 2 (dois) pesquisadores de um dos grupos que compõem o CCA. Por meio da análise da estruturação grupal e da produção discursiva observada, procuramos mostrar a relação entre experiências, papéis sociais e ações cognitivo-discursivas. Destacamos, nas análises discursivas das entrevistas, a descrição e a inter-relação de recategorizações, processos metadiscursivos (segmentos e metaformulações) e evocações/transformações em/de frames. Notamos, ao analisar esses construtos cognitivo-discursivos, que eles se coordenam entre si de acordo com a experiência social e os papéis grupais dos sujeitos entrevistados. Identificamos e problematizamos, no tratamento dessas divergências e convergências discursivas, a noção de perspectiva (GRAUMANN, 1993; VAN DIJK, 2012 [2008]), definida como o lugar social marcado por determinadas experiências sociais. Concluímos que o aparente "fosso" existente entre as experiências sociais de cada sujeito não impede a coesão entre suas perspectivações discursivas. Nossos resultados ajudam a compreender melhor a dinâmica de perspectivas sociais dentro do grupo e a refletir sobre os estudos de organização grupal / Abstract: In this thesis, we discuss the role of cognitive/discursive processes in the identification of social perspectives in a group composed by researchers and aphasics. We aim at analyzing cognitive/discursive actions carried out in thematic interviews (ALBERTI, 2004) with participants of the Social Center of Aphasics (CCA), located at the Language Studies Institute (IEL/UNICAMP) ¿ a center for social interactions among aphasic and non-aphasic people. We elaborated an interview about the group organization and the collective/individual experiences carried out in CCA. We interviewed 2 (two) aphasics and 2 (two) researchers, members of CCA. By means of the analysis of the group structure and the discursive production, we show the relationship among social experiences, roles and cognitive/discursive actions. Results show the inter-relation among re-categorizations, meta-discursive processes and activation of frames. In analyzing these processes, we noted that there is coordination among them, according to social experiences and roles that participants have in the group. We identified and questioned the notions of social perspective (GRAUMANN, 1993; VAN DIJK, 2012 [2008]) defined as the social location marked by some social experiences. We concluded that the existent "gap" among the involved social experiences does not prevent the cohesion among perspectives. Our results collaborate to the study of group organization and the dynamics of social perspectives in group / Mestrado / Linguistica / Mestre em Linguística

Orientador: Edwiges Maria Morato / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Estudos da Linguagem / Made available in DSpace on 2018-08-23T18:55:27Z (GMT). No. of bitstreams: 1 Olsen-Rodrigues_Janaina_M.pdf: 1548985 bytes, checksum: d36032a0d2d5e7ea261ceb62c9a29894 (MD5) Previous issue date: 2013 / Resumo: Com base em uma abordagem sociocognitiva da linguagem, esta pesquisa investiga e analisa as construções referenciais da hiperonímia, bem como as estratégias de ativação dos objetos de discurso a elas subjacentes, formuladas por afásicos, por indivíduos com Doença de Alzheimer em fase inicial e por indivíduos pertencentes a casos não patológicos. Como instrumento metodológico, adota-se um protocolo composto por dezoito conjuntos de três co-hipônimos cada. Tal ferramenta tem como intuito focalizar e delimitar o trabalho linguístico e sociocognitivo, assim como o percurso enunciativo realizado pelos sujeitos ao categorizar. Em consonância com esse protocolo, foi elaborado um comando que propõe aos participantes a atividade de agrupar os itens lexicais (co-hipônimos) em um mesmo "conjunto", sendo essa a noção que ampara o conceito da relação semântica de hiperonímia. A análise das estratégias empregadas pelos participantes dos casos patológicos, tal como a comparação entre elas e aquelas mobilizadas por indivíduos não afásicos e não Alzheimer, constitui a investigação científica pretendida propriamente dita. Tendo como principal motivação a necessidade de um melhor entendimento acerca das relações entre linguagem e cognição, a pesquisa fundamenta-se, do ponto de vista teóricometodológico, em três domínios: o da Semântica, o da Linguística Textual e o da Neurolinguística. Partindo da tentativa de categorizar e culminando na elaboração de enquadres cognitivos, apreciações, hiperônimos com diferentes graus de prototipicidade, holônimos, etc., os participantes revelam uma pluralidade de referentes, e explicitam o papel dos objetos de discurso como elementos dinâmicos, que são inseridos, mantidos, identificados, retomados, construindo ou reconstruindo, por esta via, os sentidos no curso da progressão textual (KOCH & MARCUSCHI, 1998; KOCH, 2002). Assim, observando como os indivíduos constroem seus agrupamentos e os representam, podemos inferir as motivações que subjazem a esse procedimento de categorização / Abstract: From a socio-cognitive approach to language, this research aims to investigate and to analyze the constructions of reference through hyperonymy formulated by individuals with aphasia, with Alzheimer's disease (in its early stages) and by individuals without any linguistic disorders from co-hyponyms presented to them. As a methodological resource, we adopt a protocol composed of eighteen sets of three co-hyponyms each. This tool has the intention to focus on and define the social cognitive and linguistic work, as well as the enunciative path used by the subjects in order to categorize. In line with this protocol, we designed a command that proposes to the participants the activity of grouping lexical items (co-hyponyms) in the same "set". This notion of grouping supports the concept of the studied semantic relationship: hyperonymy. The analysis of these strategies and also the comparison among them and those used by the control group (non-aphasic and non- Alzheimer individuals) compose the scientific investigation itself. Taking as main motivation the necessity of a better understanding of the relations among brain, language and cognition, the theoretical and methodological underpinnings of this research rest on works from three sound domains: Semantics, Textual Linguistics and Neurolinguistics. From the attempt to categorize and culminating in the development of cognitive framings, appraisals, hyperonyms with different degrees of prototypicality, holonyms, etc., participants reveal a plurality of referents, and demonstrate the role of objects of discourse as dynamic elements that are inserted, maintained, identified, building or rebuilding, in this way, the meanings in the course of textual progression (KOCH & MARCUSCHI, 1998; KOCH, 2002). Thus, by observing how individuals construct their groups and represent them, we can infer the underlying motivations in this categorization process / Mestrado / Linguistica / Mestra em Linguística

I den här uppsatsen presenteras en undersökning om konstruktioner av kön i debatten om hen. Syftet är att utifrån ett queerteoretiskt perspektiv ta reda på hur kön och hen konstrueras i en internetbaserad debatt om ordet. Undersökningen baseras på kommentarerna under ett blogginlägg om ordet hen. Det kvalitativa analysverktyget som används är membership categorization analysis (MCA). Resultatet visar att de personer som är emot användning av ordet hen är under föreställningen att ordet används för att avköna samhället, medan förespråkarna anser att ordets främsta användningsområde är som ett komplement till pronomenen han och hon. Kön konstrueras på ett tydligt sätt av motståndarna till hen, medan förespråkarna till ordet anstränger sig för att inte konstruera kvinnor och män. När kön konstrueras handlar det nästan uteslutande om konstruktion och kategorisering av kvinnor, konstruktion av män lämnas utanför debatten. Den heterosexuella matrisen och tvåkönsmodellen verkar tydligt i konstruktionen av hen och av kön och det finns en tydlig motsättning mellan motståndarna och förespråkarna till hen, då motståndarna ofta har ett essentialistiskt synsätt på kön medan förespråkarna har ett socialkonstruktivistiskt.

This software engineering thesis addresses three main issues. When creating the risk documents for this master thesis project, we became even more aware of the problems with categorization and formulation of risk statements and the scope is now focusing on categorization and formulation as a necessity for Experience based Risk Management (EbRM). The EbRM process is the foundation of the thesis and the categorisation and formulation parts had to be solved before implementing the EbRM model. To give the reader a notion about the background of this work, a brief introduction to the Experience based Risk Management model is given in the thesis. The thesis is based on literature studies, experiences and experiments. The formulation system is gathered from the Software Engineering Institute (SEI) and is called the CTC-format (Condition, Transition, Consequence). This format allows you to separate conditions and consequence of the risk and thereby provides you with easier categorisation and understandability. The categorisation system used is the SEI Taxonomy Based Categorisation (TBC). A categorisation system built as a search tree where each leaf represents a rather narrow risk domain. In order to evaluate those two different systems we performed an experiment showing that the combination thereof gave a much higher match in sorting risks between different groups. The conclusions of this work are that the TBC in connection with the CTC structure forms a very good basis for risk management when it comes to categorisation and formulation. In addition to properly formulated and tagged names and a thorough process when identifying and documenting risks, the risk management will be facilitated by using our conclusions in further risk management. Oral information must as well be on a sufficient level to gain full benefits from a risk management process.

Context: Security Requirements engineering is necessary to achieve secure software systems. Many techniques and approaches have been proposed to elicit security requirements in the initial phases of development. With the growing importance of security and immense increase in security breaches over the past few years, researchers and practitioners have been striving to achieve a mature process of coping with security requirements. Much of the activities in this regard are seen in academia but industry still seems to be lacking in giving the required importance to security requirements engineering. That is why, security requirements engineering is still not always considered as a central part of requirements engineering. This study is targeted to bridge this gap between academia and industry in terms of security requirements engineering and to provide a concrete approach to efficiently elicit and specify security requirements. The Misuse case technique is proposed for this purpose. However it lacks in providing guidelines for enabling scalable use. This limitation has been addressed to achieve a mature process of security requirements elicitation. Objectives: In this study, we propose a framework to elicit security requirements early in the software development using misuse case technique. Objective is to make misuse case technique scalable and applicable to the real-world projects. The proposed framework was presented to two representatives from the Swedish Armed Forces (SWAF). The feedback received from the representatives was utilized to refine, update and finalize the framework. Methods: The study involved a systematic review to gain an insight of the academic perspective in the area of study. Document extraction was adopted to observe the industrial trends in the said subject. These were the software requirements specification documents of the real-world systems. Document extraction was supported with informed brainstorming because the study revolved around misuse case technique and informed brainstorming is considered to be the most suitable technique for this purpose. A workshop was conducted with two representatives of Swedish Armed Forces followed by two subsequent asynchronous communication rounds and a facilitated session to get feedback about the proposed solution. This feedback was utilized to refine, update and finalize the proposed solution. Results: The results of the systematic review were organized in tabular forms for a clear understanding and easy analysis. A security requirements categorization was obtained as a result which was finalized after an initial validation with the help of real-world projects. Furthermore, a framework was proposed utilizing this categorization to address the limitations of misuse case technique. The framework was created and refined through workshop and different communication rounds with representatives of SWAF. Their feedback was used as input to further improve the usefulness and usability aspects of the framework. Conclusions: The significance of security requirements engineering is undisputedly accepted both in academia and industry. However, the area is not a subject of practice in industrial projects. The reasons include lack of mature processes as well as expensive and time consuming solutions. Lack of empirical evidences adds to the problems. The conducted study and proposed process of dealing with this issue is considered as a one step forward towards addressing the challenges. / Säkerhet Kravhantering är nödvändigt för att uppnå säkra programvarusystem. Många tekniker och metoder har föreslagits för att framkalla säkerhetskraven i de inledande faserna i utvecklingen. Med den växande betydelsen av säkerhet och enorma ökning av brott mot säkerheten under de senaste åren har forskare och praktiker strävat efter att uppnå en mogen process för att klara säkerhetskraven. Mycket av verksamheten i detta avseende ses i den akademiska världen, men industrin fortfarande tycks saknas i att ge den nödvändiga betydelse för säkerheten kravhantering. Därför är säkerheten kravhantering fortfarande inte alltid som en central del av kravhantering. Denna studie är inriktad att överbrygga denna klyfta mellan akademi och näringsliv när det gäller säkerhet kravhantering och att ge en konkret strategi för att effektivt få fram och specificera säkerhetskrav. Missbruk fallet tekniken föreslås för detta ändamål. Men det saknar i att ge riktlinjer för att möjliggöra skalbar användning. Denna begränsning har åtgärdats för att uppnå en mogen process av säkerhetskrav elicitation. / +46 (0) 735 84 12 97, +46 (0) 760 60 96 55

The aim of this study was to find out whether Europeans and North Americans differ as to what they consider to be best examples of four categories; namely vehicles, clothes, vegetables, and furniture. I compared the two continents with each other and tried to find out to what extent the cultural differences really influence the best examples chosen by the research participants. Further, I briefly compared the prototypes with European females and males and North American females and males and tried to point out some differences between the two genders. Moreover I tried to connect the differences to cultural and gender related factors. The results show the existence of some good and some bad examples that were the same no matter if we looked at the European list or the North American one. However, as we have found out through our research there seem to be strong cultural reasons for the best examples the participants chose. It is a natural behavior to choose prototypes of categories that are well known by the research participants. The best known items are those which are present in the lives of the participants. So, for example riding a bicycle does not seem to be very common among people in North America. They consider bicycle only a lower average example for the category vehicles, whereas Europeans for example seem to use bicycles much more often. They place it on rank four out of 17. People seem to choose things they know or are interested in.

Using annual reports of Austrian listed corporations between 1990 and 2005, we analyze how corporations theorize their social and societal responsibilities. We empirically illustrate that these organizations not only evoke several distinct domains of corporate responsibility, but also assign themselves and others specific positions in the social matrix of relevancy and power - which in turn gives rise to a distinct pattern on the field level. We discuss various features and implications of what we describe as a politicization of individual corporations at the price of a relocation of politics away from recognized and firmly institutionalized arenas of collective interest representation (i.e., the polity) as well as a broad-scale de-politicization of society.

In my thesis, I explore how advanced robotic technologies affect human society and my particular concern centers on investigating the boundaries between actual humans and artificial beings. Taking Steven Spielberg’s film Artificial Intelligence (2001) and Philip K. Dick’s novel Do Androids Dream of Electric Sheep (1968) as my primary sources, I illustrate how humans are experiencing dehumanization whereas artificial beings are acting much more like humans by analyzing the main characters and events that depicted in both sources. Further on, based on Nick Haslam’s theory of two main forms of dehumanization (animalistic dehumanization and mechanistic dehumanization), I discuss the interrelationships between social categorization, empathy, alienation and dehumanization by comparing actual humans and artificial beings as counter-parts. According to the descriptions of the strained relationship between these two parties, I argue that the rigid social hierarchies set foundation for dehumanization and the characteristics that define a human being, such as humanity is not a trait that only exits in humans. It can be both gained and lost.

Après avoir partagé les politiques culturelles majoritairement actives en trois modèles et modes afin dedéfinir ce qu'est La Culture, je travaille à une catégorisation de l'objet de vulgarisation de cette Culture, à savoir l'Art, et à une cartographie de son ou ses sujets, à savoir le spectateur et sa place, pour finalement essayer de proposer un mode d'action culturelle nommé non plus politique culturelle mais politique artistique. De La Critique à La Critique de la Critique de la Critique, en passant par La Critique de la Critique, je cherche à sortir de la confortable définition polysémique régnant sur le territoire culturel et à la Tour des Arts, et ce dans le projet utopique de transfiguration de leurs architectures. / The Great Culture and the Great Public as mythical as Alexander or From the art of taking distance. What is art ? Verfremdung → to make stranger. While cultural exhibition tends to be events that are most often apprehended through their relation to the Great Public – sometimes legitimating, sometimes repulsive,sometimes motive and often simultaneously recipient, adjuvant, opponent, expected, persona non grata andalmost always juge-baromètre arpenteur, supposedly outside of concern of that event – we will try to bringout from these different policies (that pay close attention to numerical parameters) the complicities and contradictions between artistic/cultural practices and democratic practice. Can we still bet on subvertissant power of cultural production as ideological enhancer and democratical antioxidant whereas the culturallandscape seems to be shaped by a festivisation oxymoronique, a nivellement-exacerbant ?