Dimensionality reduction (DR) techniques help analysts to understand patterns in high-dimensional spaces. These techniques, often represented by scatter plots, are employed in diverse science domains and facilitate similarity analysis among clusters and data samples. For datasets containing many granularities or when analysis follows the information visualization mantra, hierarchical DR techniques are the most suitable approach since they present major structures beforehand and details on demand. This work presents HUMAP, a novel hierarchical dimensionality reduction technique designed to be flexible on preserving local and global structures and preserve the mental map throughout hierarchical exploration. We provide empirical evidence of our technique's superiority compared with current hierarchical approaches and show a case study applying HUMAP for dataset labelling.

Dimensionality Reduction (DR) scatterplot layouts have become a ubiquitous visualization tool for analyzing multidimensional datasets. Despite their popularity, such scatterplots suffer from occlusion, especially when informative glyphs are used to represent data instances, potentially obfuscating critical information for the analysis under execution. Different strategies have been devised to address this issue, either producing overlap-free layouts that lack the powerful capabilities of contemporary DR techniques in uncovering interesting data patterns or eliminating overlaps as a post-processing strategy. Despite the good results of post-processing techniques, most of the best methods typically expand or distort the scatterplot area, thus reducing glyphs' size (sometimes) to unreadable dimensions, defeating the purpose of removing overlaps. This article presents Distance Grid (DGrid), a novel post-processing strategy to remove overlaps from DR layouts that faithfully preserves the original layout's characteristics and bounds the minimum glyph sizes. We show that DGrid surpasses the state-of-the-art in overlap removal (through an extensive comparative evaluation considering multiple different metrics) while also being one of the fastest techniques, especially for large datasets. A user study with 51 participants also shows that DGrid is consistently ranked among the top techniques for preserving the original scatterplots' visual characteristics and the aesthetics of the final results.

Accessibility reviews collected from app stores may contain valuable information for improving apps accessibility. Recent studies have presented insightful information on accessibility reviews, but they were based on small datasets and focused on general accessibility concerns. In this paper, we analyzed accessibility reviews that report issues affecting users with visual disabilities or conditions. Such reviews were identified based on selection criteria applied over 179,519,598 reviews of popular apps on the Google Play Store. Our results show that only 0,003% of user reviews mention visual disabilities or conditions; accessibility reviews are associated with 36 visual disabilities or eye conditions; many users do not give precise feedback and refer to their disability using generic terms; accessibility reviews can be grouped into general topics of concerns related to different types of disabilities; and positive reviews are generally associated with high scores and negative feedback with lower scores.

Automated prognosis has been a topic of intense research. Many works have sought to learn from Electronic Health Records using Recurrent Neural Networks that, despite promising results, have been overcome by novel techniques. We introduce APEHR, a Transformer approach that leverages medical prognosis using the latest technology Neural Network Transformer, which has demonstrated superior results in problems whose data is organized in sequential fashion. We contribute with an innovative problem modeling along with a detailed discussion of how Transformers can be used in the medical domain. Our results demonstrate a prognostic performance that surpasses previous works by at least 6% for metric Recall@k in the public dataset MIMIC-III.

Well-written code (which meets standards, conventions, and quality specifications) is often associated with the programmer's experience, which is why companies have been looking for increasingly qualified professionals. Codes at odds with these standards tend to be complex and poorly written and are generally difficult to understand. Consequently, the activities inherent to it become costly. Although well defined and consolidated in the industry, the concepts of code quality taught in the academy are insufficient to enable students to meet market expectations. Research indicates that graduates do not feel prepared to enter the labor market, facing difficulties when competing for the best job opportunities. This difficulty highlights a gap between industry and academia. Researchers have identified the gap and have proposed improvements to the teaching-learning process based on using the concepts and tools widely used in the software industry in an academic environment. When analyzing educational institutions' programming disciplines, it is possible to notice a mismatch between programming disciplines and code quality disciplines. In this scenario, this project aims to propose improvements on how knowledge of programming and quality is evolved, proposing an approach that uses software quality as a subsidy for teaching programming, providing the teacher with guidelines for teaching programming focusing on internal quality source code. We idealized an innovative practice that brought programming, usually focused on the execution of algorithms, within the teaching of code quality, focused on software engineering principles. The use of code inspection tools allowed direct teaching according to the class's needs, introducing a guided content-based internal quality teaching in programming disciplines without generating extra work for the teacher. In addition to the improvements for the teacher, we are convinced that from the student's perspective, we were able to motivate them to learn to program not only concerned with the execution results but also how the solution is developed.

Multidimensional projection techniques have been widely used to visually explore datasets due to their ability to generate representations that preserve similarity relations of data points into lower dimensional spaces. To evaluate if the embedded space reflects high-dimensional structures, measures are usually employed to return a quality score of the whole projection. In contrast to this idea, we evaluate the embedded layouts by assessing each class of the datasets at a time by using well-known quality measures. In addition, we propose assessing multidimensional projection techniques using ROC curves. Experimental results on two datasets show that our approach can be useful to discover how classes interact each other by using different visualization techniques and how close-related they are without thoroughly exploring the layouts. ROC curves proved to be a good measure for analyzing projection techniques and can give highly valuable feedback to users when exploring multidimensional data.

Computer Science teaching materials are biased towards concepts and theoretical aspects. One may consider it difficult to relate concepts to concrete problems. Consequently, it increases the chances of a student not recognizing the relevance of the subject, becoming unmotivated and unprepared to solve practical problems or coping with workplace needs after college. This paper shows the use of social media data as an alternative to minimize the skill gap between what the student learns in college and the skills required in the workplace. The proposed strategy consists of extracting topics from Stack Overflow questions to identify concepts generally unknown or misunderstood and concepts that their practical application represents a challenge. The concepts covered in Stack Overflow questions provide strong cues about how professors and instructors can improve teaching material with useful content for their students, as well as increase their motivation -- since the teaching material becomes clearly related to practical issues in Computer Science. We show, by an example, how to use the proposed strategy to improve teaching material from the generated topics. Also, we demonstrate that the application of topic modeling in Stack Overflow content is promising as a professor support tool to enhance teaching material relevance in Computer Science courses and diminish the college-workplace skill gap.

Software Quality Assurance is a sub-process that ensures that developed software meets and complies with defined or standardized quality specifications. Focusing on source code, there are characteristics that can be used to evaluate the quality. Introductory courses must encourage freshmen students to improve internal quality of their source code, but only as sophomore they have contact with Software Engineering concepts, including Quality Assurance. In this paper we present a tool to source code quality evaluation aimed at supporting students to improve their source code and, consequently, their programming skills. The proposed tool uses quality reports (available to professional environment integrate with software repositories) to analyze students' source code and provide a feedback about the student coding. The proposed tool run locally, with few computational resources. In addition, we proposed the methodology to use the proposed tool: it consists of challenging students to perform a set of maintenance tasks in a controlled environment. We prepared a source code by introducing common defects, what decreases the quality of source code, and ask to students to perform maintenance tasks in order to both eliminate the introduced defects and introduce new features. After each modification, the students must evaluate their code using the proposed tool to obtain a feedback about quality of source code. To evaluate the approach and the tool, we created a survey and applied to students and the teacher. As a result, we show the benefits of using the proposed tool to both teachers and students perspectives. The results are positive to enhance the teaching-learning Software Quality Assurance to Software Engineering students.

Theoretical Computer Science area (TCS) stands out by being an important study field, and it is composed by Formal Languages and Automata Theory (FLA), Computer Science Theory (CST), and Theory of Compilers (TC). This area is responsible for introducing the beginnings of the Computer Science through formalisms - which represent a set of methods, techniques, or rules that describe the solution to a problem with restrictions - and it has a substantial impact on the student's knowledge. Computer science theory is based on the understanding of computability and techniques to solve challenges, and to improve the teaching-learning process used to introduce these concepts we proposed a Combined Methodology for Theoretical Computing (CMTC). Our methodology is based on formalism development to ground the knowledge acquired during classes of FLA, CST, and TC, where students are introduced to Theoretical Computing during one year and a half. In each course, we applied the same methodology where each student used data structures, computer graphics, and algorithms to solve problems. We address this methodology to understand how much the incomprehension of formalisms is influenced by new concepts and its abstractions. Against this background, we demonstrate that the that CMTC has the aim to build knowledge and make the new concepts and formalisms concrete. Our results are based on statistical analysis from students' grades, where we could observe among other results, the correlation between the practical activities and the conceptual knowledge.

Several multidimensional projection techniques have been proposed in literature over the last years. The quality of those techniques can be evaluated based on the dimensionality reduction or the clusters quality. The first evaluation aim to verify if the similarities from multidimensional space are preserved in projected space. While the second evaluation aim to verify if instances from a same class are placed in a same cluster in projected space. Respectively, Stress and Silhouette Coefficient are the main measures to quality evaluations. In this paper we present two new approaches -- named Simplified Stress and Simplified Silhouette Coefficient -- to speed up the computation of measures, enabling a faster evaluation of multidimensional projection techniques and feature spaces. We present experiments showing the high correlation between results obtained using original approaches and results obtained with those proposed in this paper. In addition, we show how to use Simplified Silhouette Coefficient to perform a fast feature space evaluation and selection.

Visualization applications can be performed on distinct platforms, such as mobile devices and multi-projection systems. Each platform offers specific features to provide further data understanding, and a system that integrates these platforms in a complementary manner is a real challenge. In this paper, we present an immersive and interactive visualization system that aims to explore data from relational databases using 3D graphs representations, where multiple simultaneous users can visualize and interact with the data through a multi-projection system and mobile devices. A single visualization application was created for both platforms using the Unity game engine, and an Unity external package for Virtual Reality applications development, that supports multi-projection system over a PC cluster and passive stereoscopy. Our visualization system aims to provide the users a better data understanding using a 3-screens multi-projection system as data overview, and mobile devices as display and interaction device for navigation and additional information visualization. We also introduce an user case, where the visualization system is used in order to support developers regarding structural problems in a large relational database.

Visualization techniques have been widely used in dataset exploration. A common strategy is to employ different techniques to facilitate the exploratory process, enabling different perspectives from the same dataset. In this case, a coordination mechanism aids the user in the context changing among different views. However, to keep track of the highlighted data elements among multiple views is an unclear task. To reach a better exploration by using distinct visualization techniques and graphical representations, some approaches have adopted the strategy of combining different techniques in a single view, creating a Hybrid Visualization which can avoid the use of coordination. This paper proposes a new Hybrid Visualization approach that integrates Multidimensional Projection and Parallel Coordinates to display the instances relationship and attributes behaviour in a single view. As presented in this paper, the applications of this approach enable a better feature space exploration, aiding users to understand why instances from the same or distinct classes are grouped.

The number of images available has grown over the years, as well as the number of techniques to aid to organizing and retrieving from image collections. Techniques and systems have been proposed to recover images based on query, in which an image (or words) is used as input parameter and a list of similar images (or images with related text content) is recovered. However, understanding how the retrieved images are related to each other remains as a problem. This paper proposes an approach based on multidimensional visualization and coordination techniques to show the relationship from retrieved images. In this approach, coordination techniques are employed to perform image retrieval methods and highlight the results in visual representations, showing how retrieved images are relate. To evaluate our proposal image collections with and without textual annotations related to each image were used, and also image retrieval mechanisms based on distance, topic and semantic to retrieve images from distinct and multimodal datasets.

The wide availability of database systems and low cost of hardware allow enterprises and researchers the opportunity to store large data collections. The challenge then became the understanding of these data. To overcome this problem Information Visualization (IV) techniques have been employed to amplify the human cognitive ability through graphical data representations, that show properties and relationships from these data. This work presents an approach to overcome the visual scalability by using a Multi-projection system, allowing the exploration of large datasets. Additionally, this approach allows collaborative interaction and exploration by using mobile devices like tablets and smart phones.

Visualization techniques have proved to be valuable tools to support textual data exploration. Dimensionality reduction techniques have been widely used to produce visual representation of document collections. Focusing on multidimensional projection techniques, good visual results are produced depending on how representative terms to discriminate the documents are chosen to compose the vector space model (VSM). To define a good VSM it is necessary to apply filters during the preprocessing in order to eliminate terms using their frequency. For that, the user must evaluate the term frequency histogram based on his/her expertise in the text subject and decide the threshold value for frequency cut. Usually it is a trial and error approach that requires the user to verify the quality of visual representation after each trial. In this paper, we propose an automatic approach that applies the Otsu's Threshold Selection Method for computing a threshold using a term frequency histogram. We conducted experiments that have shown our approach generates visual representations as good as those generated with a threshold obtained by trial and error approach. The contribution of our approach is that users with non expertise are able to generate good visual representations and the time to get a good threshold is decreased.

Multiple views of data sets from the same domain can support to discover unforeseen associations among data elements, but requires users to interact with them. The coordination mechanism must relate elements across multiple views. The mapping among data elements are constrained by using data attributes, and such mapping influences on how multiple views are coordinated. We propose the application of ontology to link data elements based on semantic for specific context. Representing the underlying data into ontology, semantic representation to create the mappings can benefit exploratory visualization. In this paper we show how to use ontology on coordinating multiple views, the initial results using document collections are presented and discussed, in comparison with traditional techniques.

Exploratory tasks supported by visualization are usually improved by Coordinated and Multiple Views (CMV) of the data under study. Several coordination techniques have been proposed in the literature, resulting in a diversity of tools to generate mappings among the multiple views. These mappings can be highly dynamic, and their history reveals the settings employed in the multiple exploratory tasks conducted in a discovery process. Several solutions have been proposed to help users to recover the steps performed in exploratory tasks, but little support is found for registering the multiple coordination mappings employed. This paper provides a contribution in this direction, proposing a model for storing and recovering such mappings. We believe such a facility is an important feature of CMV systems, so that users can recover and rerun the coordinations performed when exploring their data. We present details of the proposed model and show some potential applications.

Abstract Fiber tracts detection is an increasingly common technology for diagnosis and also understanding of brain function. Although tools for tracing and presenting brain fibers are advanced, it is still difficult for physicians or students to explore the dataset in 3D due to their intricate topology. In this work we present a visual exploration approach for fiber tracts data aimed at supporting exploration of such data. The work employs a local, precise and fast 2D multidimensional projection technique that allows a large number of fibers to be handled simultaneously and to select groups of bundled fibers for further exploration. In this approach, a DTI feature dataset, including curvature as well as spatial features, is projected on a 2D or 3D view. By handling groups formed in this view, exploration is linked to corresponding brain fibers in object space. The link exists in both directions and fibers selected in object space are also mapped to feature space. Our approach also allows users to modify the projection, controlling and improving, if necessary, the definition of groups of fibers for small and large datasets, due to the local nature of the projection. Compared to other related work, the method presented here is faster for creating visual representations, making it possible to explore complete sets of fibers tracts up to 250K fibers, which was not possible previously. Additionally, the ability to change configuration of the feature space representation adds a high degree of flexibility to the process.

Abstract Multidimensional projection has emerged as an important visualization tool in applications involving the visual analysis of high‐dimensional data. However, high precision projection methods are either computationally expensive or not flexible enough to enable feedback from user interaction into the projection process. A built‐in mechanism that dynamically adapts the projection based on direct user intervention would make the technique more useful for a larger range of applications and data sets. In this paper we propose the Piecewise Laplacian‐based Projection (PLP), a novel multidimensional projection technique, that, due to the local nature of its formulation, enables a versatile mechanism to interact with projected data and to allow interactive changes to alter the projection map dynamically, a capability unique of this technique. We exploit the flexibility provided by PLP in two interactive projection‐based applications, one designed to organize pictures visually and another to build music playlists. These applications illustrate the usefulness of PLP in handling high‐dimensional data in a flexible and highly visual way. We also compare PLP with the currently most promising projections in terms of precision and speed, showing that it performs very well also according to these quality criteria.

Classification is an important task for computer-aided diagnosis systems (CADs). However, many classifiers may not perform well, presenting poor generalization and high computational cost, especially when dealing with high-dimensional datasets. Thus, feature selection can greatly mitigate these problems. In this paper, we propose two filter-based feature selection algorithms that calculate the simplified silhouette statistic as evaluation function: the silhouette-based greedy search (SiGS) and the silhouette-based genetic algorithm search (SiGAS). Silhouette statistic is used to guide the search for features that provide better class separability. Experiments performed on three datasets have shown that the SiGAS algorithm overcomes traditional filter algorithms, such as CFS, FCBF and reliefF. It also outperforms a similar algorithm, kNNGAS, based on genetic algorithm that minimizes the classification error of k-nearest neighbors. Additionally, results have shown that SiGAS produces better accuracy than SiGS.

Document interpretation is a crucial task in many visual analytics applications, made harder by the widespread availability of freely available textual files. In this paper we propose an approach based on topic detection coupled with multiple coordinated views to assist analysis of time varying document collections. Given multiple document maps built from a set of text files, we define a strategy to support users locating the evolution of topics addressed by the documents, along various time steps. The approach is supported by a new algorithm for topic extraction from texts, also introduced. Finally, we show several examples illustrating how the proposed strategy may be applied in the analysis of document collections.

Multiple Views have been put forward as an alternative to assist exploration of evolving phenomena or associations between distinct data sets or distinct presentations of a single data set. Coordinating between views is a challenge that must be met to improve visualization support for exploratory tasks. This is particularly true for high-dimensional data, such as document collections. We introduce a coordination framework for multiple views of document collections created using projections and point placement visualizations. Coordination can occur between different views of a single data set or between views of multiple data sets. Multiple coordinations are also admitted. Three new types of coordination are presented to illustrate the framework; these have been implemented in a multipurpose multi-dimensional visualization system called PEx (Projection Explorer).