Version-controlled documents, such as Wikipedia or program codes in Subversion, demands a novel methodology to be analyzed efficiently. The documents are continually edited by one or more authors in contrast of the case of static documents. These collaborative processses make traditional methodologies to be ineffective, yet needs for efficient methodologies are rapidly developing. This paper proposes two new models based on Local Space-time Smoothing (LSS) which captures important revision patterns while Cumulative Revision Map (CRM) tracks word insertions and deletions in particular positions of a document. These two methods enable us to understand and visualize the revision patterns intuitively and efficiently. Synthetic data and real-world data are used to demonstrate its applicability.

A Visualização Computacional trata de técnicas para representar e interagir graficamente com dados complexos, em geral de alta dimensionalidade. Dados de alta dimensionalidade são caracterizados por pontos representados em espaços vetoriais de alta dimensão, cada coordenada representando um atributo do vetor. Num grande número de aplicações da visualização multidimensional uma medida de similaridade existe entre esses vetores. Técnicas de projeção multidimensional podem ser utilizadas para posicionamento desses dados num plano de forma a facilitar a interpretação das relações de similaridade. Entretanto alguns problemas dessas técnicas comprometem a interpretação dos resultados obtidos. Este trabalho identifica esses problemas e propõe, uma técnica para posicionar os pontos no plano, através da formação de árvores filogenéticas a partir de relações de similaridade. Em geral árvores filogenéticas são utilizadas para codificação de relações de ancestralidade. Um algoritmo de geração e um algoritmo de traçado dessas árvores foram implementados no contexto do sistema PEx (Projection Explorer) e a solução é comparada com a funcionalidade das projeções na interpretação de dados multidimensionais em geral e, em particular, na representação de coleções de documentos, uma aplicação bastante estratégica da visualização computacional e da mineração visual de dados / Computational Visualization is concerned with graphical representation and exploration of complex data, usually bearing high dimensionality. Multidimensional data are characterized by points represented in vector spaces of many dimensions, each coordinate representing an attribute of the vector. In many applications a similarity measure can be found to highlight relationships of proximity between the vectors. In this environment projection techniques offer an alternative to ease interpretation coded by the similarity measures through proximity on the display. They do so by positioning the points on a bidimensional plane. Projection techniques are very useful to display and interact with data, but present some drawbacks that in some cases compromise the interpretation of certain features in data sets. This work discusses such problems and proposes, as an alternative to represent similarity relationships and to provide point placement on a plane, the use of phylogenetic trees, a representation typically employed to represent ancestrality relationships. An algorithm for generation and an algorithm for drawing such trees were implemented in a system called Projection Explorer. The approach is compared to that of multidimensional projections for multidimensional data in general and, in particular, for document data sets, an strategic application for multidimensional visualizations, since text can be represented and interpreted as multi-dimensional entities

A Visualização Computacional trata de técnicas para representar e interagir graficamente com dados complexos, em geral de alta dimensionalidade. Dados de alta dimensionalidade são caracterizados por pontos representados em espaços vetoriais de alta dimensão, cada coordenada representando um atributo do vetor. Num grande número de aplicações da visualização multidimensional uma medida de similaridade existe entre esses vetores. Técnicas de projeção multidimensional podem ser utilizadas para posicionamento desses dados num plano de forma a facilitar a interpretação das relações de similaridade. Entretanto alguns problemas dessas técnicas comprometem a interpretação dos resultados obtidos. Este trabalho identifica esses problemas e propõe, uma técnica para posicionar os pontos no plano, através da formação de árvores filogenéticas a partir de relações de similaridade. Em geral árvores filogenéticas são utilizadas para codificação de relações de ancestralidade. Um algoritmo de geração e um algoritmo de traçado dessas árvores foram implementados no contexto do sistema PEx (Projection Explorer) e a solução é comparada com a funcionalidade das projeções na interpretação de dados multidimensionais em geral e, em particular, na representação de coleções de documentos, uma aplicação bastante estratégica da visualização computacional e da mineração visual de dados / Computational Visualization is concerned with graphical representation and exploration of complex data, usually bearing high dimensionality. Multidimensional data are characterized by points represented in vector spaces of many dimensions, each coordinate representing an attribute of the vector. In many applications a similarity measure can be found to highlight relationships of proximity between the vectors. In this environment projection techniques offer an alternative to ease interpretation coded by the similarity measures through proximity on the display. They do so by positioning the points on a bidimensional plane. Projection techniques are very useful to display and interact with data, but present some drawbacks that in some cases compromise the interpretation of certain features in data sets. This work discusses such problems and proposes, as an alternative to represent similarity relationships and to provide point placement on a plane, the use of phylogenetic trees, a representation typically employed to represent ancestrality relationships. An algorithm for generation and an algorithm for drawing such trees were implemented in a system called Projection Explorer. The approach is compared to that of multidimensional projections for multidimensional data in general and, in particular, for document data sets, an strategic application for multidimensional visualizations, since text can be represented and interpreted as multi-dimensional entities

Mapas de documentos são representações visuais que permitem analisar de forma eficiente diversas relações entre documentos de uma coleção. Técnicas de projeção multidimensional podem ser empregadas para criar mapas que refletem a similaridade de conteúdo, favorecendo a identificação de agrupamentos com conteúdo similar. Este trabalho aborda uma evolução do arcabouço genérico oferecido pelas projeções multidimensionais para apoiar a análise interativa de documentos textuais, implementado na plataforma PEx. Foram propostas e implementadas técnicas que permitem ao usuário interagir com o mapa de documentos utilizando tópicos extraídos do próprio corpus. Assim a representação visual pode gradualmente evoluir para refletir melhor os interesses do usuário, e apoiá-lo de maneira mais efetiva em tarefas exploratórias. A interação foi avaliada utilizando uma técnica de inspeção de usabilidade, que visa identificar os principais problemas enfrentados pelos usuários ao interagir com as funcionalidades desenvolvidas. Adicionalmente, a utilidade das funcionalidades foi avaliada pela condução de dois estudos de caso, em que foram definidas tarefas a serem conduzidas pelo usuário sobre os mapas de documentos. Os resultados mostram que com o auxílio das visualizações foi possível conduzir as tarefas satisfatoriamente, permitindo manipular de forma eficiente milhares de documentos sem a necessidade de ler individualmente cada texto / Content-based document maps are visualizations that help users to identify and explore relationships among documents in a collection. Multidimensional projection techniques have been employed to create similaritybased maps that can help identifying documents of similar content. This work aims to enhance the generic framework offered by the multidimensional projection techniques in the PEx visualization platform to support interactive analysis of textual data. Several interaction functions and visual representations have been proposed and implemented that allow users to interact with document maps aided by topics automatically extracted from the corpus. By exploring the topics and maps in an integrated manner, users can refine and evolve the visual representations gradually to better reflect their needs and interests, enhancing support to exploratory tasks. The proposed interaction functions were evaluated employing a usability inspection technique, seeking to detect interface problems. Moreover, two illustrative case studies were conducted to evaluate the usefulness of the proposed interactions, based on typical user tasks defined over different document collections. They illustrate how the developed visualizations can assist the proposed tasks, allowing users to interactively explore large document corpora and refine document maps

Mapas de documentos são representações visuais que permitem analisar de forma eficiente diversas relações entre documentos de uma coleção. Técnicas de projeção multidimensional podem ser empregadas para criar mapas que refletem a similaridade de conteúdo, favorecendo a identificação de agrupamentos com conteúdo similar. Este trabalho aborda uma evolução do arcabouço genérico oferecido pelas projeções multidimensionais para apoiar a análise interativa de documentos textuais, implementado na plataforma PEx. Foram propostas e implementadas técnicas que permitem ao usuário interagir com o mapa de documentos utilizando tópicos extraídos do próprio corpus. Assim a representação visual pode gradualmente evoluir para refletir melhor os interesses do usuário, e apoiá-lo de maneira mais efetiva em tarefas exploratórias. A interação foi avaliada utilizando uma técnica de inspeção de usabilidade, que visa identificar os principais problemas enfrentados pelos usuários ao interagir com as funcionalidades desenvolvidas. Adicionalmente, a utilidade das funcionalidades foi avaliada pela condução de dois estudos de caso, em que foram definidas tarefas a serem conduzidas pelo usuário sobre os mapas de documentos. Os resultados mostram que com o auxílio das visualizações foi possível conduzir as tarefas satisfatoriamente, permitindo manipular de forma eficiente milhares de documentos sem a necessidade de ler individualmente cada texto / Content-based document maps are visualizations that help users to identify and explore relationships among documents in a collection. Multidimensional projection techniques have been employed to create similaritybased maps that can help identifying documents of similar content. This work aims to enhance the generic framework offered by the multidimensional projection techniques in the PEx visualization platform to support interactive analysis of textual data. Several interaction functions and visual representations have been proposed and implemented that allow users to interact with document maps aided by topics automatically extracted from the corpus. By exploring the topics and maps in an integrated manner, users can refine and evolve the visual representations gradually to better reflect their needs and interests, enhancing support to exploratory tasks. The proposed interaction functions were evaluated employing a usability inspection technique, seeking to detect interface problems. Moreover, two illustrative case studies were conducted to evaluate the usefulness of the proposed interactions, based on typical user tasks defined over different document collections. They illustrate how the developed visualizations can assist the proposed tasks, allowing users to interactively explore large document corpora and refine document maps

De nombreux travaux ont montré que, à l’origine des limitations de l’interaction bureautique, il y a un manque à la fois d’espace moteur et d’espace visuel.Cette thèse explore des moyens pour optimiser l’usage de ces espaces.D’une part, à partir du constat que l’on contrôle la position et l’orientation d’un objet par un geste naturel, cette thèse étudie les bénéfices que peut offrir une souris dotée d’un capteur de rotation. Cette < souris rotative > permet à l’utilisateur de maîtriser avec aisance trois variables continues d’une tâche informatique. Un état de l’art présente des aspects perceptifs et moteurs des actions en question et les particularités ergonomiques et techniques d’un tel dispositif. Deux techniques d’interaction - visant à faciliter des tâches métier récurrentes - sont proposées :le réglage < quasi-intégral > et la < palette satellitaire >. Par ailleurs, une évaluation expérimentale compare la performance d’une souris rotative avec celle d’une souris traditionnelle.D’autre part, ce travail se penche sur les questions de la visualisation de documents dans le contexte des bibliothèques numériques. D’abord, il examine l’apport et la faisabilité technique de l’utilisation d’un dispositif d’affichage immersif pour la navigation dans un catalogue de titres virtuel. Puis, afin de faciliter l’inspection massive d’un lot de pages numérisées, il avance des techniques de visualisation zoomables et multi-focales. Ces dernières permettent, dans une recherche d’anomalies, de saisir vite les caractéristiques visuelles de quelques centaines de pages. Et cela grâce à un va-et-vient entre la vue d’ensemble et la navigation panoramique des détails. / Past research has suggested that among the reasons for the limitations of present desktop interaction style is the lack of both motor and visual space. The goal ofthis thesis is to optimize the use of such spaces. Based on the fact that one can control an object’s position and orientation through a natural movement, the first main contributioin of this thesis is to explorethe advantages of enhancing the sensing of the standard mouse througha rotation sensor. This < rotary mouse > allows one to easily control three continuous variables of a computer task. A survey presents the perceptual and motorissues of some rotary manipulations and also the technical and ergonomic requirements of such device. Two interaction techniques, aimed to simplify repetitive tasks, are proposed : the < nearly-integral selection > and the < satellite palette >.Furthermore, an experimental evaluation compares the performance of the rotarymouse with that of a standard one. The other main contribution of this work is to investigate document visualization issues in the context of digital libraries. First, it analyses the advantages and the technical feasibility of integrating an immersive display to an interface aimed to support navigation in a virtual catalog. Second, in order to inspect the quality of a batch of digitized pages, it explores some zoomable and multi-focal visualization techniques. The overview and the panoramic detail browsing enabled by such techniques try to help users, which have to identify the flaws resulted from the digitization process, to quickly grasp the visual characteristics of a large set of pages.

De nombreux travaux ont montré que, à l’origine des limitations de l’interaction bureautique, il y a un manque à la fois d’espace moteur et d’espace visuel.Cette thèse explore des moyens pour optimiser l’usage de ces espaces.D’une part, à partir du constat que l’on contrôle la position et l’orientation d’un objet par un geste naturel, cette thèse étudie les bénéfices que peut offrir une souris dotée d’un capteur de rotation. Cette < souris rotative > permet à l’utilisateur de maîtriser avec aisance trois variables continues d’une tâche informatique. Un état de l’art présente des aspects perceptifs et moteurs des actions en question et les particularités ergonomiques et techniques d’un tel dispositif. Deux techniques d’interaction - visant à faciliter des tâches métier récurrentes - sont proposées :le réglage < quasi-intégral > et la < palette satellitaire >. Par ailleurs, une évaluation expérimentale compare la performance d’une souris rotative avec celle d’une souris traditionnelle.D’autre part, ce travail se penche sur les questions de la visualisation de documents dans le contexte des bibliothèques numériques. D’abord, il examine l’apport et la faisabilité technique de l’utilisation d’un dispositif d’affichage immersif pour la navigation dans un catalogue de titres virtuel. Puis, afin de faciliter l’inspection massive d’un lot de pages numérisées, il avance des techniques de visualisation zoomables et multi-focales. Ces dernières permettent, dans une recherche d’anomalies, de saisir vite les caractéristiques visuelles de quelques centaines de pages. Et cela grâce à un va-et-vient entre la vue d’ensemble et la navigation panoramique des détails. / Past research has suggested that among the reasons for the limitations of present desktop interaction style is the lack of both motor and visual space. The goal ofthis thesis is to optimize the use of such spaces. Based on the fact that one can control an object’s position and orientation through a natural movement, the first main contributioin of this thesis is to explorethe advantages of enhancing the sensing of the standard mouse througha rotation sensor. This < rotary mouse > allows one to easily control three continuous variables of a computer task. A survey presents the perceptual and motorissues of some rotary manipulations and also the technical and ergonomic requirements of such device. Two interaction techniques, aimed to simplify repetitive tasks, are proposed : the < nearly-integral selection > and the < satellite palette >.Furthermore, an experimental evaluation compares the performance of the rotarymouse with that of a standard one. The other main contribution of this work is to investigate document visualization issues in the context of digital libraries. First, it analyses the advantages and the technical feasibility of integrating an immersive display to an interface aimed to support navigation in a virtual catalog. Second, in order to inspect the quality of a batch of digitized pages, it explores some zoomable and multi-focal visualization techniques. The overview and the panoramic detail browsing enabled by such techniques try to help users, which have to identify the flaws resulted from the digitization process, to quickly grasp the visual characteristics of a large set of pages.