Pattern extraction algorithms are enabling insights into the ever-growing amount of today's datasets by translating reoccurring data properties into compact representations. Yet, a practical problem arises: With increasing data volumes and complexity also the number of patterns increases, leaving the analyst with a vast result space. Current algorithmic and especially visualization approaches often fail to answer central overview questions essential for a comprehensive understanding of pattern distributions and support, their quality, and relevance to the analysis task. To address these challenges, we contribute a visual analytics pipeline targeted on the pattern-driven exploration of result spaces in a semi- automatic fashion. Specifically, we combine image feature analysis and unsupervised learning to partition the pattern space into interpretable, coherent chunks, which should be given priority in a subsequent in-depth analysis. In our analysis scenarios, no ground-truth is given. Thus, we employ and evaluate novel quality metrics derived from the distance distributions of our image feature vectors and the derived cluster model to guide the feature selection process. We visualize our results interactively, allowing the user to drill down from overview to detail into the pattern space and demonstrate our techniques in two case studies on Earth observation and biomedical genomic data.
Research into how virtual reality (VR) can be a beneficial technology for new and emerging large, complex data visualizations for data scientists is ongoing. In this paper, we evaluate three-dimensional VR navigation technique for data visualizations and test their effectiveness with a large graph visualization. We evaluate two prominent navigation techniques employed in VR (Teleportation and One-Handed Flying) against two less common methods (Two-Handed Flying and Worlds In Miniature) and evaluate their performance and effectiveness through a series of tasks. We found Steering Patterns (One-Handed Flying and Two-Handed Flying) to be faster and preferred by participants for completing searching tasks in comparision to Teleportation. Worlds-In-Miniature was the least physically demanding of the navigations, and was preferred by participants for tasks that required an overview of the graph such as triangle counting.
Axes are the main components of coordinate systems representations. They play a critical role for the visual analysis of multi-dimensional data. However their representation seems to have always be considered self evident, with oriented lines crossing at an origin, completed with labels such as ticks and names. Such classical representation show limits when it comes 3D visualization and immersive analytic (IA), mainly because orthogonal projection of points on linear axes is hard in a 3d environment, and because the user can move therefore the axes can get out of his field of view. In this paper we propose a task-based definition of axes and coordinate systems representation, as well as a tentative design space for coordinates systems representation in immersion. We also present an exploratory user study we carried out to compare three grid-based representations of coordinate systems for multidimensional data analysis with 3D scatterplots.
Building management systems (BMS) provide monitoring and control of most large-building assets (heating, ventilation, air conditioning, lighting, security systems, and so on). With the recent advancement of the Internet of Things and data management systems, BMS must gather and manage increasingly detailed data coming from a greater number and diversity of sources. The availability of such data should help building managers optimise the energy consumption of buildings. However, current BMS don't allow efficient visualisation of such data, which means that even if the data is available, it is not used to its full potential. In this paper, we describe a prototype BMS interface providing interactive visualisations of traditional building data (temperature, energy consumption), as well as more novel data (comfort feedback from occupants and live occupancy). We evaluate this prototype by first showing how it could be used to plan a long- term energy saving strategy, and then in a feedback session involving facility managers at a university.
Social media allows citizens, corporations, and authorities to create, post, and exchange information. The study of its dynamics will enable analysts to understand user activities and social group characteristics such as connectedness, geospatial distribution, and temporal behavior. In this context, social media bubbles can be defined as social groups that exhibit certain biases in social media. These biases strongly depend on the dimensions selected in the analysis, for example, topic affinity, credibility, sentiment, and geographic distribution. In this paper, we present SocialOcean, a visual analytics system that allows for the investigation of social media bubbles. There exists a large body of research in social sciences which identifies important dimensions of social media bubbles (SMBs). While such dimensions have been studied separately, and also some of them in combination, it is still an open question which dimensions play the most important role in defining SMBs. Since the concept of SMBs is fairly recent, there are many unknowns regarding their characterization. We investigate the thematic and spatiotemporal characteristics of SMBs and present a visual analytics system to address questions such as: What are the most important dimensions that characterize SMBs? and How SMBs embody in the presence of specific events that resonate with them? We illustrate our approach using three different real scenarios related to the single event of Boston Marathon Bombing, and political news about Global Warming. We perform an expert evaluation, analyze the experts' feedback, and present the lessons learned.
The analysis of invasive team sports often concentrates on cooperative and competitive aspects of collective movement behavior. A main goal is the identification and explanation of strategies, and eventually the development of new strategies. In visual sports analytics, a range of different visual-interactive analysis techniques have been proposed, e.g., based on visualization using for example trajectories, graphs, heatmaps, and animations. Identifying suitable visualizations for a specific situation is key to a successful analysis. Existing systems enable the interactive selection of different visualization facets to support the analysis process. However, an interactive selection of appropriate visualizations is a difficult, complex, and time-consuming task. In this paper, we propose a four-step analytics conceptual workflow for an automatic selection of appropriate views for key situations in soccer games. Our concept covers classification, specification, explanation, and alteration of match situations, effectively enabling the analysts to focus on important game situations and the determination of alternative moves. Combining abstract visualizations with real world video recordings by Immersive Visual Analytics and descriptive storylines, we support domain experts in understanding key situations. We demonstrate the usefulness of our proposed conceptual workflow via two proofs of concept and evaluate our system by comparing our results to manual video annotations by domain experts. Initial expert feedback shows that our proposed concept improves the understanding of competitive sports and leads to a more efficient data analysis.
Visualization is widely accepted as an effective medium to communicate complex data to a human observer. To do this effectively, visualizations have to be carefully designed to achieve a certain intent. Visualization guidelines are proposed by the academic research community and practitioners to facilitate effective visualization design. A few guidelines have been received a fair amount of attention, and effort has been made to study, discuss, validate, falsify, adopt, adapt, or extend them. However, many guidelines have not received adequate exposure or have not had the opportunities to undergone a similar level of scrutiny. When some of these guidelines managed to emerge or resurface, it is often not clear about their scientific rationale and the state of play in their validation. In this paper, we juxtapose the development and consumption of visualization guidelines with that of consumer products. We outline a conceptual model for a Visualization Guidelines Supply Chain, VISupply. It describes an idealized loop of actions for formulating, curating, using, and improving guidelines systematically. By enabling an ecosystem for visualization guidelines, the community can collectively optimize these guidelines and adopt them with confidence in a given context. We examine the current and potential roles of different stakeholders in this ecosystem.
Bipartite graphs are typically visualized using linked lists or matrices. However, these classic visualization techniques do not scale well with the number of nodes. Biclustering has been used to aggregate edges, but not to create linked lists with thousands of nodes. In this paper, we present a new casual exploration interface for large, weighted bipartite graphs, which allows for multi-scale exploration through hierarchical aggregation of nodes and edges using biclustering in linked lists. We demonstrate the usefulness of the technique using two data sets: a database of media advertising expenses of public authorities and author-keyword co-occurrences from the IEEE Visualization Publication collection. Through an insight-based study with lay users, we show that the biclustering interface leads to longer exploration times, more insights, and more unexpected findings than a baseline interface using only filtering. However, users also perceive the biclustering interface as more complex.
In recent years, interactive visualization to analyze text documents has gained an impressive momentum. This is not surprising considering the fast increase of electronically available textual documents of various kinds. These include, for example, patents, scholarly documents, social media messages, and many other sources that contain valuable knowledge and insights for many stakeholders. Interactive text visualization turned out to be an important means for exploring and gaining insights into complex and often large document collections. An established visualization strategy to represent such collections is using projection-based techniques that visualize documents as glyphs in a 2D view aiming to reflect the semantic similarity of documents by the proximity of their placement. Static labels have been suggested to characterize the overall topics contained in the projected data to improve the effectiveness of such visualization techniques. Other approaches employ magic lenses that enable users to explore the 2D spatialization freely on various granularity levels. In this work, we propose a visual exploration approach that combines cluster-based labeling of projected documents with an interaction concept for magic lens techniques. We offer a set of novel interactive features to support a smooth transition between static labels and the magic lens approach while exploiting the different levels of visual abstraction of both techniques without introducing additional clutter through overdraw. Finally, we provide insights gained from a preliminary user study and present the benefits of our approach.
We present LTMA, a Layered Topic Matching approach for the unsupervised comparative analysis of topic modeling results. Due to the vast number of available modeling algorithms, an efficient and effective comparison of their results is detrimental to a data- and task-driven selection of a model. LTMA automates this comparative analysis by providing topic matching based on two layers (document-overlap and keyword-similarity), creating a novel topic-match data structure. This data structure builds a basis for model exploration and optimization, thus, allowing for an efficient evaluation of their performance in the context of a given type of text data and task. This is especially important for text types where an annotated gold standard dataset is not readily available and, therefore, quantitative evaluation methods are not applicable. We confirm the usefulness of our technique based on three use cases, namely: (1) the automatic comparative evaluation of topic models, (2) the visual exploration of topic modeling differences, and (3) the optimization of topic modeling results through combining matches.
We describe a system designed to process, analyze and visualize academic data, from research papers and research proposals to list of courses taught, consulting, internal and external service. This can be helpful in identifying experts in a given field for future collaborations, as well as in putting together strong multi-disciplinary teams to apply for future research funding. Our REMatch system aims to support such tasks by leveraging natural language processing, machine learning, and interactive visualization. Specifically, REMatch provides a functional system that implements in-the-browser, map-based interactive navigation of a large underlying network, supporting semantic zooming, panning, searching, and map overlays. A prototype of the system is evaluated with a small-scale case study.
We address the problem of visualizing and interacting with large multi-dimensional time- series data. We propose a visual analytics system and approach which aims to visualize, analyze, present and enable exploration of large temporal datasets. Our approach consists of three main stages which are preprocessing, dimensionality reduction, and visual exploration. It assists with finding the interesting features in the data which are often obscured in the line chart because of the visual compression that is required to render the large dataset to screen. Our approach helps to obtain an overview of the entire dataset and track changes over time. It enables the user to detect clusters and outliers and observe the transitions between data. The juxtaposed views are used to visualize and interact both with raw time series data and projected data. Different time series datasets are deployed on our system, and we demonstrate the utility and evaluate the results using a case study with two different datasets which show the effectiveness of our system.
Recent advances in virtual reality technologies enable high-fidelity exploration of data in immersive environments. This is advantageous for professional applications of high-dimensional datasets (such as geo-temporal narratives), as we can leverage all three spatial axes while immersing the user in the information itself. Geo-temporal narratives tell a story of entities, their movements, and as a result, their potential relationships, thereby defining the who, what, where, and when that define a story; everything except the why. This paper describes an immersive virtual reality system we have developed to convey these narratives, specifically focusing on the law enforcement domain. The system lets users not only view who was where and when, but also view explicit and implicit relationships between entities, repeated visits to recurring locations, as well as the crucial descriptive information supporting the why. We present the results of an expert review of the system from federal law enforcement and defence agencies that validate our approach.
Multiple Coordinated Views (MCV) has been widely used in visualization. This work explores Multiple Coordinated Spaces (MCS), a 3D version of MCV, in order to integrate various 2D displays in a large physical environment as a unified analysis workspace. Built upon the rich background of distributed and embodied cognition, MCS supports interactive analysis in a connected, distributed set of subspaces. For MCS, we have developed visualization and interactive techniques for coordinating augmented reality devices together with classical WIMP GUIs systems. We also demonstrate the usages of MCS using a multivariate, geo-spatial biodiversity application. The major advantage of MCS is a flexible coordination framework for creating new immersive analytics methods by mixing visualizations from different devices, and mixing physical and virtual operations from different environments.
Exploratory visual analysis is an iterative process, where analysts often start from an overview of the data. Subsequently, they often pursue different hypotheses through multiple rounds of interaction and analysis. Commercial visualization packages support mostly a model with a single analysis path, where the system view represents only the final state of the users' current analysis. In this paper, we investigate the benefit of using multiple workspaces to support alternative analyses, enabling users to create different workspaces to pursue multiple analysis paths at the same time. We implemented a prototype for multiple workspaces using a multi-tab design in a visual analytics system. The results of our user studies show that multiple workspaces: enable analysts to work on concurrent tasks, work well for organizing an analysis, and make it easy to revisit previous parts of their work.
Despite the resurgence of virtual reality (VR), the primary method of interacting with the environment is using generic controllers. Given the often-purpose-built nature of applications within VR, this is surprising, as despite the effort put into the design of the application itself, the same attention is not paid to the input control. This is despite the advantages that tangible interfaces have for user understanding, especially in the context of visualization, where user understanding is paramount. This paper presents the adaptation of a previous 2D temporal-geospatial visualization into VR, and more importantly, describes the development of a novel 8DOF Tangible User Interface developed to support the exploration of that data. For our application, this centers around the exploration of geospatial data to explore colocation and divergence of entities, but could easily be extended to other domains. We present our novel controller as an example of the benefits of the utilization of purpose built physical controllers as a first-tier method of enabling immersive analytics. We describe the immersive system and controller, followed by an example use case and other applications encouraging further development of novel tangibles as a key component of immersive data analytics.
The following topics are dealt with: data visualisation; data analysis; interactive systems; pattern clustering; virtual reality; Big Data; graph theory; text analysis; user interfaces; sport.
Large, high-resolution displays (LHRDs) have been shown to enable increased productivity over conventional monitors. Previous work has identified the benefits of LHRDs for Visual Analytics tasks, where the user is analyzing complex data sets. However, LHRDs are fundamentally different from desktop and mobile computing environments, presenting some unique usability challenges and opportunities, and need to be better understood. There is thus a need for additional studies to analyze the impact of LHRD size and display resolution on content spatialization strategies and Visual Analytics task performance. We present the results of two studies of the effects of physical display size and resolution on analytical task successes and also analyze how participants spatially cluster visual content in different display conditions. Overall, we found that navigation technique preferences differ significantly among users, that the wide range of observed spatialization types suggest several different analysis techniques are adopted, and that display size affects clustering task performance whereas display resolution does not.
Personal networks formed within scientific communities and the collaborations they yield are one of the driving forces behind innovation and new discoveries. Luckily, successful collaboration produces analyzable data points in the form of publications that allow us to learn and understand some of the connections and collaborative structures in a scientific community. Co-author information is one important aspect of this, and various solutions to the fundamental visualization problems of co-author graphs exist. In this work, we introduce ColTop, a multi- level, interactive graph visualization system that allows users to effectively analyze publication data. It combines coauthor information with other meta-data and information extracted from textual content to support comprehensive analyses. ColTop includes a novel, heuristics-based approach to create a meaningful abstraction of co-author networks, and enriches them with topic information. To demonstrate the applicability of our approach, we discuss an example analysis scenario based on a practical data set.
We present HoloBee, a system for visual analytics of bee drift data using a head-mounted augmented reality interface (Microsoft HoloLens). HoloBee facilitates interactive manipulation and exploration of the bee activity data on 3D geo-spatial terrains to gain insights into the drifting behaviour of honey bees in their natural habitat. We conducted a user study to compare the task performance and user experience of HoloBee with a conventional desktop application. The results show that HoloBee allows for solving tasks in a similar time and accuracy compared to the desktop interface, despite participants' having significantly more experience interacting with a desktop interface. We received positive feedback from a subset of participants on the natural and intuitive interactions that HoloBee provides, while others preferred using the desktop application. We discuss the implications and limitations of these results and provide guidance for future work in this direction.
