Deep reinforcement learning is a dynamic field that has been successfully applied in various engineering and scientific disciplines. However, like many areas of automated learning, it presents a significant challenge: understanding its models, which can hinder human trust in the decisions made by these algorithms. To address this issue, we conducted a focus group and interviews with experts in machine learning and reinforcement learning to gain insights into the perceptions and preferences surrounding interpretability techniques and tools in reinforcement learning systems. The discussion of the focus groups used an interactive dashboard to simplify the analysis of reinforcement learning environments. The focus group results showed many problems to tackle with the current approaches regarding RL agent training, like data manipulation, environment representation and reward misinterpretation. The focus group discussion also presented some key features of a visualization approach to interpretability in RL: contextual presentation of information, integration with existing pipelines, and ease of distribution.

Situated visualisation (SV), which embeds data representations directly within the context to which they refer, offers a paradigm for augmenting perception, understanding, and decision-making in real-world environments. Inspired by comic-books and previous work on visualisation, this article presents the relevance of SV in the form of superpowers. Each superpower is explained through the lens of augmented reality (AR). By mapping these capabilities, the article positions SV not just as a technical advancement, but as a new frontier in human-centred design, inviting researchers to rethink the role of visualisation in a world where data and reality increasingly coexist. Therefore, the paper also offers guidelines for its effective design and implementation. Finally, a structured evaluation questionnaire (the VIEWUS), specifically designed for AR-based Situated visualisation systems, is proposed. VIEWUS integrates key factors such as usability, engagement, situational awareness, visual fatigue, and workload, combining and adapting items from established instruments. It allows for a comprehensive assessment using a 5-point Likert scale and supports iterative improvement of SV systems based on user feedback.

This study aims to review academic research publications on techniques that deal with the shortcomings of classic information visualization charts. This review used snowballing to capture the academic papers and applied a filter to focus on articles that developed a new technique, resulting in 28 articles. We proposed some research questions and a taxonomy to classify the papers and conduct a detailed analysis of each study. In this way, the study identified one solution proposal per article that makes use of one or more of the ten (10) visual shortcomings mitigation techniques found for each of the five (5) problems related to Visual shortcomings that affect one of the five (5) information visualization techniques. As a result, we found solutions and mapped them to each of the identified visual shortcomings, including overlap, outliers, categorical occlusion, misuse of space, and non-perceptible elements. These solutions help to minimize the problem as a whole or in part, making it easier for the user to understand the information in the graphs. This article characterizes an initial effort to study academic papers that solved or mitigated shortcomings in classical information visualization charts.

This paper presents an Augmented Reality (AR) application based on Situated Visualization (SV), aimed at increasing awareness of air pollution. Given that technological progress must be achieved through meticulous evaluation processes, the article also presents the assessment of the stated application. Therefore, this article begins by examining how AR applications are appraised. Building upon this review, and focusing particularly on those where SV is provided, a discussion is presented regarding the points that must be carefully considered when evaluating this type of AR application, as well as the factors that need to be measured, along with the most popular methods for evaluating these factors. However, it is noted that if a simultaneous evaluation of all the factors that matter to SV is intended, the number of questions to be asked could be significant. Therefore, a preliminary approach to enable such an appraisal is proposed and used in the mentioned application to demonstrate the merits of the proposed Questionnaire.

Data coding and decoding are essential steps in communicating and interpreting information. These encodings are basic units of data representation, which can be combined and are generally organized by a layout of a data visualization technique, a common scenario for digital data visualizations. Technological advances in manipulating materials allowed us to represent data physically, creating an opportunity to use other senses, in addition to vision and hearing, to encode and decode data, such as touch, smell, and taste. In everyday life, touch is already widely used to decode information with few values using vibrations, such as turning devices on/off, communicating alerts on devices, and providing feedback in interaction in video games, among others. Still, it is generally not used to decode a larger set of values. Therefore, this article aims to investigate the encoding and decoding of data through tactile vibrations, identifying the minimum and maximum limits of tactile perception without causing discomfort and minimum intervals that differentiate two consecutive vibration values. This preliminary information allowed us to understand how to map data using vibrotactile scales. The data are mapped according to their value, changing the vibration frequency of the motors. Two stages of testing were carried out. The first aims to find the limits of tactile perception and create the vibrotactile scale, and the second aims to perform data analysis tasks, such as comparison, ordering, and identification. The results demonstrate that users considered find minimum and maximum vibration values tasks easy, while clustering and sorting tasks were classified as the most complex.

This study presents Glyphforge, an automated visual encoding method that makes use of visual variables that partially overlap. The heuristics utilized to optimize the efficiency of the partial overlap factor were derived from studies centered around visual perception assessments. By combining preprocessing and feature selection techniques, this approach aims to identify and retain a maximum of five key features from a given dataset. A visualization technique utilizing layered multidimensional data glyphs was implemented for the purpose of the visual encoding. This tool considers some use cases for the end users, like manipulation of columns and layers, and inverting the dimensions for the glyph layers.

The use of Pervasive Augmented Reality (AR) in industrial settings, particularly in the context of Industry 4.0, is a growing area of research. Industry 4.0 is characterized by the integration of advanced technologies, including AR, to enable smart, connected, and autonomous systems in manufacturing and other industries. This study focuses on two systems that use Pervasive AR for logistics and data monitoring on assembly lines. Such systems were designed using cooperative Human-Centered Design (HCD) methodologies in collaboration with partners from the industry sector. We report lessons learned and guidelines on the development of AR systems for industrial environments. In addition, a user study based on a data monitoring laboratory realistic procedure with 8 participants is described. The study compares the usability and effectiveness of a dashboard on Pervasive AR and a web tablet. Overall, results emphasize that AR was better suited when requiring spatial awareness and interventions, while traditional web tablet was better for tasks that require more interaction with the content and faster data monitoring. These findings had implications for the design of industrial tools, such as prototype improvements and idealize new methods, which can guide future efforts to develop and deploy AR applications.

Augmented Reality (AR) is a key technology in the transition to Industry 4.0 and smart manufacturing, gaining reputation in a wide range of industrial fields One promising application is in another field of Industry 4.0, the data monitoring, where AR can be used to visualize and interact with complex and big data in real time, potentially improving the efficiency and accuracy of decision-making In this work, we propose a Pervasive AR tool for data monitoring, developed in collaboration with industry partners. An web application of the same data monitoring function is also created for comparison purposes A Human-Centered Design (HCD) methodology was used to identify the needs and requirements of industrial analysts, which led to the development of these systems Preliminary user studies, with 17 participants having distinct levels of expertise in industry, data analysis and computer science, were conducted to collect opinions and suggestions for improvements, as well as, quantitative data regarding the technologies considered A succeeding user study was then prepared, in which 12 participants used two conditions (C1 - Hands-free Pervasive AR tool for HMDs and C2 - Web tool for tablet devices) to fulfill a set of data monitoring tasks in an industry assembly line The result of these studies confirm the potential of Pervasive AR for data monitoring, as it engages users, promote environmental awareness, contextualizes data and allows fast self-localization.

Recent literature reviews point out that active data physicalizations still lack evaluation scenarios. These evaluations mainly focus on the perception of visual and physical characteristics of physicalizations and users' performance of analysis tasks. This paper addresses a proposed accuracy assessment of data physicalization for data representation. The data physicalization used is a physical and dynamic bar chart using 3D printed models (3D bars) and LED strips inside the bars, both controlled by microcontrollers (Arduino), being possible to represent different datasets. The accuracy assessment of this physicalization is based on the height and color of the 3D bars. For this, distance sensors measure the actual height values of the bars, and a computer vision camera/algorithm collects and estimates the RGB value of the bars. From original and measured values datasets, bar charts images are built to perform a similarity comparison between the images. The similarity metric is used as an accuracy metric of the bar heights to represent data and obtained as initial results value higher than 96%. Furthermore, an evaluation scenario of the distance between the reference and measured colors 2000) classifies most of them as having proper perception. Additionally, the accuracy loss was evaluated on data physicalization in a scenario of continuous use (ten configurations without any calibration), and the variation of errors was around 2%.

The value of good data visualization has already been shown in several scenarios. Still, it is not always easy to obtain it, as it depends on factors such as the dataset, the amount of data, task types, the user profile, the type of interaction, etc. To mitigate the challenges addressed, automated or semi-automated systems have been proposed, emphasizing rule-based/heuristic approaches and machine-learning models. However, many of these applications require specialized knowledge and present results (data visualizations) that are not flexible for customization. Papers have highlighted the ease of tools like ChatGPT in creating various tasks, including creating data charts. This facility, in addition to the intelligent computational model involved, is also due to the expressiveness used in the requests to execute the tasks by the users since these tools use Natural Language Interfaces. Despite adopting these tools overgrowing in different scenarios of society, studies on the best way to use them, integrate them into existing processes, or evaluative studies on their effectiveness or efficiency are still incipient. Thus, this paper will evaluate the workload for creating data visualization using ChatGPT 3.5. For assessment, the NASA Task Load Index (Nasa TLX) methodology was applied, and users with experience creating data visualization created two proposed scenarios. The preliminary results showed high temporal and mental demand, mainly due to the vocabulary used and the completeness of the user instructions. The average time to create and perform InfoVis tasks in two proposed evaluation scenarios was 33 and 44 minutes, and 14 queries were applied on average for both scenarios. The direct consequence was that the users have redone the requests and improved the instructions at each new iteration, and all users completed the proposed tasks.

In recent years, the physicalization area has expanded rapidly and has been utilized in numerous contexts, mostly in big open public areas or data-related places. Most data physicalizations have visual restrictions and no interaction. Hence, multiple technologies, including augmented reality (AR), have been applied to physical visualizations to address the above-mentioned issues. This research proposes dynamic data physicalization using AR virtual content (visual data items). The dynamic bar chart allows the configuration of numerical data to height, categorical data to color, and categorical data to the x-axis. The mobile augmented reality (MAR) application performs some Infovis tasks, such as settings, filters, details on demand, etc. A cloud server selects data, calculates visual elements or additional visualizations, calculates scales for physicalizing data, and enables the communication between the MAR application and dynamic physicalization. Lastly, dynamic and augmented data physicalization characteristics and scenarios are shown.

The process of creating physical data visualizations is not a trivial task. In general, it may require skills from the user in information visualization, tangible interaction, 3D modeling, fabrication of physical objects, etc. In addition, few works have presented computational support to the entire digital and physical rendering pipeline of data visualization, characterizing many steps of this process as manual. From this context, this work presents a process that facilitates the generation of physical data visualization. Besides that, It allows one to define which type of physical visualization to create, among passive, rearrangeable, and dynamic physicalization types. Finally, the pipelines for each physicalization type are presented, with scenarios including physical bar charts, stacked bar charts, and grouped bar charts.

This paper aims to present the design process of dynamic data physicalization for bar charts and their variants, simple bars, stacked bars, and clustered bars. The physical artifact has 12 bars that are automatically configured according to the dataset by electromechanical components controlled by an Arduino board. The models of the printed 3D parts, the electromechanical components used and their connection scheme, the design aspects to set up the dynamic physical visualization according to the data, how the users choose the dataset and the kind of physical chart, usage scenarios with different types of charts and physicalization parameters are presented in detail. Finally, some strengths and difficulties in creating dynamic physical visualizations are highlighted and future works are proposed.

To support the nuances of collaborative work, many researchers have been exploring the field of Augmented Reality (AR), aiming to assist in co-located or remote scenarios. Solutions using AR allow taking advantage from seamless integration of virtual objects and real-world objects, thus providing collaborators with a shared understanding or common ground environment. However, most of the research efforts, so far, have been devoted to experiment with technology and mature methods to support its design and development. Therefore, it is now time to understand where the field stands and how well can it address collaborative work with AR, to better characterize and evaluate the collaboration process. In this article, we perform an analysis of the different dimensions that should be taken into account when analysing the contributions of AR to the collaborative work effort. Then, we bring these dimensions forward into a conceptual framework and propose an extended human-centered taxonomy for the categorization of the main features of Collaborative AR. Our goal is to foster harmonization of perspectives for the field, which may help create a common ground for systematization and discussion. We hope to influence and improve how research in this field is reported by providing a structured list of the defining characteristics. Finally, some examples of the use of the taxonomy are presented to show how it can serve to gather information for characterizing AR-supported collaborative work, and illustrate its potential as the grounds to elicit further studies.

Even with the growth of digital data sources and support for creating visualizations, much information disseminated in chart format is still static, whether printed or digital. Some common problems of visualization components can be easily corrected on the chart design using a charting tool, but in the real world, mainly in the printed media, it is not a simple task. We investigate a method to reconstruct visualizations in real-time and in a real-world environment using Augmented Reality. We propose a prototype that interacts with extracted data from a bitmap chart in the real world, and we evaluated it with usability experts. The results show a high degree of satisfaction on many factors, mainly filter usage and valuable feedback about interaction.

This work aims to review the academic literature on information visualization techniques used in dashboards applied to strategic business or management decision-making in different application areas. The review used the snowballing method to obtain academic works and applied a filter to focus on papers published in journals or conference proceedings, reaching 44 papers. We propose four research questions and one taxonomy to classify the works, carry out the analyses, and later a research agenda to address the identified gaps. For instance, this review revealed a lack of academic papers that discuss the subject involving state-of-the-art information visualization and machine learning techniques. This paper represents our initial effort to examine academic works that use information visualization techniques and theory applied to dashboards that support decision-making in business or management areas.

Augmented Reality (AR) has been explored with the objective to assist in scenarios of co-located or remote collaboration. To help understand how well collaborative work can be addressed with AR, it is important to foster harmonization of perspectives and create a common ground for systematization and discussion. In this vein, understand relationships among existing dimensions of collaboration, as well as identify research opportunities, is of paramount importance and thus tools that allow visually exploring information associated with Collaborative AR may be most valuable. In this paper, we present a first effort towards the creation of such an interactive visualization tool for exploration and analysis of collaborative AR research. It allows visualize data of selected papers organized according to a human-centered taxonomy on collaborative AR. In order to get insights into whether the structure was understood and if the representation was clear and efficient to use, we evaluated the proposed tool through a user study with 40 participants. Results suggest the tool has potential towards the creation of a shared understanding and identification of existing patterns, trends and opportunities within the field of collaborative AR.

Artificial Intelligent (AI) techniques, such as ma-chine learning (ML), have been making significant progress over the past decade. Many systems have been applied in sensitive tasks involving critical infrastructures which affect human well-being or health. Before deploying an AI system, it is necessary to validate its behavior and guarantee that it will continue to perform as expected when deployed in a real-world environment. For this reason, it is important to comprehend specific aspects of such systems. For example, understanding how neural networks produce final predictions remains a fundamental challenge. Existing work on interpreting neural network predictions for images via feature visualization often focuses on explaining predictions for neurons of one single convolutional layer. Not presenting a global perspective over the features learned by the model leads the user to miss the bigger picture. In this work we focus on providing a representation based on the structure of deep neural networks. It presents a visualization able to give the user a global perspective over the feature maps of a convolutional neural network (CNN) in a single image, revealing potential problems of the learning representations present in the network feature maps.

With vast amounts of data, comes vast numbers of problems. The process of collecting data is far from perfect, either due to human factors or technological errors, which can lead to inaccuracies and uncertainties in the data. One such issue is missing data: the absence of information. Several methods can deal with missing values, but to choose the correct approach, it is necessary to diagnose the missing data mechanisms, which describe how the distribution of missingness in a given data variable correlates to other variables. This diagnosis can be made with statistical tests or data visualization techniques. However, statistical tests provide an uncertainty estimation that is often misinterpreted, and the visualizations readily available in data analysis packages have some scalability issues, such as cognitive overload and lack of screen space. Thus, this paper proposes a visual-interactive idiom for diagnosing missing data mechanisms. The proposed solution consists of a set of visual encodings and two derived metrics that synthesizes the missing data mechanisms and the uncertainty associated with this synthesis. We present the concepts behind the visual encodings, derived metrics, and interactions of the idiom.

The decision-making process and the development of decision support systems (DSS) have been enhanced by a variety of methods originated from information science, cognitive psychology and artificial intelligence over the past years. Situated visualization (SV) is a method to present data representations in context. Its main characteristic is to display data representations near the data referent. As augmented reality (AR) is becoming more mature, affordable and widespread, using it as a tool for SV becomes feasible in several situations. In addition, it may provide a positive contribution to more effective and efficient decision-making, as the users have contextual, relevant and appropriate information to endorse their choices. As new challenges and opportunities arise, it is important to understand the relevance of intertwining these fields. Based on a literature analysis, this paper addresses and discusses current areas of application, benefits, challenges and opportunities of using SV through AR to visualize data in context and to support a decision-making process and its importance in future DSS.

The textual description of data charts is a complex task. A chart presents different visual characteristics for the information represented, which can be influenced by the technique selected and the combination of visual elements. There are crowdsourcing initiatives to create descriptions for charts available on the Web, but the descriptions can have failures, considering that they arise from the understanding of the person. In this context, methods to automatically extract data from chart images allow producing descriptions for use in these scenarios. However, there is no standard way of vocalizing the chart content. For this, the textual description must be based on a template, so that the chart can be completely understood. Thus, this paper presents templates that allow verbalizing the data extracted from vertical and grouped bar charts in an intelligible way. Evaluations were performed with users to verify the ease of understanding textual descriptions. The results showed that the proposed templates were suitable for vocalization the contents of bar charts.

The scarcity of chart images public datasets and its generators makes difficult the comparative studies between works in areas such as information visualization, usability, computer vision, and machine learning. Therefore, this paper presents a Web tool for generating data chart images called Synthetic Chart Image Generator (SCIG). The tool uses VEGA declarative language, and the chart features are fully parameterizable for every eleven different classes, as well as image quantity and its resolution. Additionally, the charts are constructed from synthetic data, randomly generated by probability distributions functions, and rendered in PNG format. Finally, this paper presents a performance test of the chart images generation.

The evaluation is an essential step of works that propose new information visualization techniques or tools. A common type is the controlled experiment in which the researchers measure the user performance to execute specific tasks using the proposed method. Furthermore, the dataset used for these tests must contain known desired features to be evaluated (e.g., level of noise, the percentage of missing values) in a controlled way. Thus, this article proposes an application to generate synthetic databases for evaluating information visualization techniques and tools. The system aims to create a dataset generator model that allows the construction of datasets with a diversity of profiles in a controlled manner. The creator of the model can save it for future experiments or updates and can export it enabling other groups to replicate the experiments easily. For a better understanding of application features and how to use it, this work also shows two use scenarios explaining the created model for each situation.

Different information visualization techniques can be found in the literature due to the quantity and variety of data stored in computational systems. In this context, the classification of chart images becomes important because it allows various types of graphs to be detected automatically in different contexts, allowing a more specific processing for each type of visualization, for example, data extraction. Several techniques of image classification can be used, where the most common are based on the extraction of features of the images, and a later classification using these features. However, one technique that has been gaining prominence in the context of image classification is the Convolutional Neural Network (CNN). This technique is based on deep learning and, in a way, encapsulates the feature extraction process. In this way, the proposal of this article is to use an architecture of a client-server based model to do the chart image classification and later data extraction from this image. The main advantage is doing the CNN processing on the server side, so the application does not rely on client device limitations. For this, an image dataset was generated from the web, and it has ten classes of graphs. From the experiments done, it was seen that the use of this technique was feasible, and modifications in the architecture can be made as a proposal to improve the accuracy of the model.

This paper presents a starting study on InfoVis interaction through mid-air gestures, using a vision based interaction device, the Leap Motion. We present the user tests conducted and the results gathered, using a 3D Scatterplot as visualization technique. These tests aim in identify and categorize issues that can compromise InfoVis mid-air gestural interaction as a whole, not focusing on the design of the developed tool. The test tasks and results are exposed and the issues found are categorized in Boundary Awareness, Granularity and Collision and Depth Perception.

This paper presents results of head tracker and speech commands interaction evaluation in information visualization context at low level task taxonomy. The head tracker is used to move a pointer and the voice commands to perform actions. In the evaluation, we used 20 public information visualization techniques, 62 simple tasks (aiming low-level interactions), and 10 participants performed all tasks. The evaluation shows results about user performance (UP) and user experience (UE) by visualization and interaction taxonomies. Finally, we highlight some good practices suggestions for design of this type of interaction.

Design and develop collaborative methods in information visualization are a current challenge, hence understand, identify and define the current progress of developed methods to collaborate in information visualization are important. The aim of this work is to review the ways and strategies of how to collaborate in information visualization applications. We surveyed published works that present applications that implements collaborative information visualization techniques and methods. We present a micro level category of identified ways of how the InfoVis applications implements collaboration.

The current Brazilian context in television audiovisual communication is in transition from analogical to digital technology. One of the main reasons for this motion towards Interactive Digital Television (iDTV) is the improvement in the user experience over this platform. Considering this motion, this work applied suitable interaction design and usability concepts from the Nielsen's heuristics [5] into a visualization prototype to the support of visual analysis of products in electronic commerce over TV (t-commerce). In this work we conducted a usability check in an existing information visualization prototype, in order establish whether or not it follows usability standards and in order to make it more usable to the user.

This paper identifies an opportunity to reduce the latency in information visualization (InfoVis) systems, exploring the parallelization of the visualization pipeline architecture. We propose a concurrent architecture where the visualization pipeline stages are modified to execute as producers and consumers threads. The threads synchronization is done by memory barriers and the data flow pass the pipeline through a unique data structure, called ring buffer, which reuses a contiguous space preallocated in memory. Two InfoVis prototypes were developed in java, the first one using sequential pipeline and the other using concurrent pipeline. The results obtained with concurrent architecture in comparison with sequential pipeline presented less execution time and memory allocation for data visualization renderization.

The Internet has played an important role as a knowledge-sharing network and in this context some service oriented architecture (SOA) applications have emerged in all kind of study fields. Therefore, this work aims the design and development of a service aggregation that will favor ubiquity and pervasiveness in data visualization applications, allowing users to build domain-specific data visualizations in an easy and intuitive way. With this kind of service, it is possible to build data visualization applications for different smart devices such as smartphones, tablets, desktop, smart TV's, etc. A Web API that supports the main functionalities of an information visualization tool in different platforms has been proposed to reaches these purposes. The REST (Representational State Transfer) Style has been employed in the service conception as the architectural communication model. Client-side and server-side applications were developed using Java with a data visualization generator engine called PRISMA.