IEEE VIS Publication Dataset

The process of learning models from raw data typically requires a substantial amount of user input during the model initialization phase. We present an assistive visualization system which greatly reduces the load on the users and makes the process of model initialization and refinement more efficient, problem-driven, and engaging. Utilizing a sequence segmentation task with a Hidden Markov Model as an example, we assign each token in the sequence a feature vector based on its various properties within the sequence. These vectors are then clustered according to similarity, generating a layout of the individual tokens in form of a node link diagram where the length of the links is determined by the feature vector similarity. Users may then tune the weights of the feature vector components to improve the segmentation, which is visualized as a better separation of the clusters. Also, as individual clusters represent different classes, the user can now work at the cluster level to define token classes, instead of labelling one entry at time. Inconsistent entries visually identify themselves by locating at the periphery of clusters, and the user then helps refine the model by resolving these inconsistencies. Our system therefore makes efficient use of the knowledge of its users, only requesting user assistance for non-trivial data items. It so allows users to visually analyse data at a higher, more abstract level, improving scalability.

This poster describes our progress in developing an interactive linear modeling system that supports the modeling approach described by Daniel and Wood. Our visual interface permits analysts to build sets of possible models and then creates appropriate visualizations to permit human-in-the-loop model comparison and selection.

In this paper, we introduce ALIDA, an Active Learning Intent Discerning Agent for visual analytics interfaces. As users interact with and explore data in a visual analytics environment they are each developing their own unique analytic process. The goal of ALIDA is to observe and record the human-computer interactions and utilize these observations as a means of supporting user exploration; ALIDA does this by using interaction to make decision about user interest. As such, ALIDA is designed to track the decision history (interactions) of a user. This history is then utilized to enhance the user's decision-making process by allowing the user to return to previously visited search states, as well as providing suggestions of other search states that may be of interest based on past exploration modalities. The agent passes these suggestions (or decisions) back to an interactive visualization prototype, and these suggestions are used to guide the user, either by suggesting searches or changes to the visualization view. Current work has tested ALIDA under the exploration of homonyms for users wishing to explore word linkages within a dictionary. Ongoing work includes using ALIDA to guide users in transfer function design for volume rendering within scientific gateways.

Co-located collaboration can be extremely valuable during complex visual analytics tasks. This paper presents an exploratory study of a system designed to support collaborative visual analysis tasks on a digital tabletop display. Fifteen participant pairs employed Cam-biera, a visual analytics system, to solve a problem involving 240 digital documents. Our analysis, supported by observations, system logs, questionnaires, and interview data, explores how pairs approached the problem around the table. We contribute a unique, rich understanding of how users worked together around the table and identify eight types of collaboration styles that can be used to identify how closely people work together while problem solving. We show how the closeness of teams' collaboration influenced how well they performed on the task overall. We further discuss the role of the tabletop for visual analytics tasks and derive novel design implications for future co-located collaborative tabletop problem solving systems.

Modern machine learning techniques provide robust approaches for data-driven modeling and critical information extraction, while human experts hold the advantage of possessing high-level intelligence and domain-specific expertise. We combine the power of the two for anomaly detection in GPS data by integrating them through a visualization and human-computer interaction interface. In this paper we introduce GPSvas (GPS Visual Analytics System), a system that detects anomalies in GPS data using the approach of visual analytics: a conditional random field (CRF) model is used as the machine learning component for anomaly detection in streaming GPS traces. A visualization component and an interactive user interface are built to visualize the data stream, display significant analysis results (i.e., anomalies or uncertain predications) and hidden information extracted by the anomaly detection model, which enable human experts to observe the real-time data behavior and gain insights into the data flow. Human experts further provide guidance to the machine learning model through the interaction tools; the learning model is then incrementally improved through an active learning procedure.

Insight Externalization (IE) refers to the process of capturing and recording the semantics of insights in decision making and problem solving. To reduce human effort, Automated Insight Externalization (AIE) is desired. Most existing IE approaches achieve automation by capturing events (e.g., clicks and key presses) or actions (e.g., panning and zooming). In this paper, we propose a novel AIE approach named Click2Annotate. It allows semi-automatic insight annotation that captures low-level analytics task results (e.g., clusters and outliers), which have higher semantic richness and abstraction levels than actions and events. Click2Annotate has two significant benefits. First, it reduces human effort required in IE and generates annotations easy to understand. Second, the rich semantic information encoded in the annotations enables various insight management activities, such as insight browsing and insight retrieval. We present a formal user study that proved this first benefit. We also illustrate the second benefit by presenting the novel insight management activities we developed based on Click2Annotate, namely scented insight browsing and faceted insight search.

The Self-Organizing Map (SOM) algorithm is a popular and widely used cluster algorithm. Its constraint to organize clusters on a grid structure makes it very amenable to visualization. On the other hand, the grid constraint may lead to reduced cluster accuracy and reliability, compared to other clustering methods not implementing this restriction. We propose a visual cluster analysis system that allows to validate the output of the SOM algorithm by comparison with alternative clustering methods. Specifically, visual mappings overlaying alternative clustering results onto the SOM are proposed. We apply our system on an example data set, and outline main analytical use cases.

We present an iterative strategy for finding a relevant subset of attributes for the purpose of classification in high-dimensional, heterogeneous data sets. The attribute subset is used for the construction of a classifier function. In order to cope with the challenge of scalability, the analysis is split into an overview of all attributes and a detailed analysis of small groups of attributes. The overview provides generic information on statistical dependencies between attributes. With this information the user can select groups of attributes and an analytical method for their detailed analysis. The detailed analysis involves the identification of redundant attributes (via classification or regression) and the creation of summarizing attributes (via clustering or dimension reduction). Our strategy does not prescribe specific analytical methods. Instead, we recursively combine the results of different methods to find or generate a subset of attributes to use for classification.

Interaction and manual manipulation have been shown in the cognitive science literature to play a critical role in problem solving. Given different types of interactions or constraints on interactions, a problem can appear to have different degrees of difficulty. While this relationship between interaction and problem solving has been well studied in the cognitive science literatures, the visual analytics community has yet to exploit this understanding for analytical problem solving. In this paper, we hypothesize that constraints on interactions and constraints encoded in visual representations can lead to strategies of varying effectiveness during problem solving. To test our hypothesis, we conducted a user study in which participants were given different levels of interaction constraints when solving a simple math game called Number Scrabble. Number Scrabble is known to have an optimal visual problem isomorph, and the goal of this study is to learn if and how the participants could derive the isomorph and to analyze the strategies that the participants utilize in solving the problem. Our results indicate that constraints on interactions do affect problem solving, and that while the optimal visual isomorph is difficult to derive, certain interaction constraints can lead to a higher chance of deriving the isomorph.

Advanced battlespace network visualization techniques are required within the modern Air Operations Center (AOC) to improve cross-domain situation awareness and to support planning and decision-making. We present a visualization toolkit to address this need that supports the integration of network health and status information and meta-information with other traditional AOC information resources and activities across air, space, and cyber domains. Applications include the development of battlespace visualization technologies that will improve warfighters' decision-making response time and provide enhanced flexibility for mission planning by efficiently revealing affordances for leveraging, disrupting, or enhancing network connectivity.

Geometric Wavelets is a new multi-scale data representation technique which is useful for a variety of applications such as data compression, interpretation and anomaly detection. We have developed an interactive visualization with multiple linked views to help users quickly explore data sets and understand this novel construction. Currently the interface is being used by applied mathematicians to view results and gain new insights, speeding methods development.

Journalists increasingly turn to social media sources such as Facebook or Twitter to support their coverage of various news events. For large-scale events such as televised debates and speeches, the amount of content on social media can easily become overwhelming, yet still contain information that may aid and augment reporting via individual content items as well as via aggregate information from the crowd's response. In this work we present a visual analytic tool, Vox Civitas, designed to help journalists and media professionals extract news value from large-scale aggregations of social media content around broadcast events. We discuss the design of the tool, present the text analysis techniques used to enable the presentation, and provide details on the visual and interaction design. We provide an exploratory evaluation based on a user study in which journalists interacted with the system to explore and report on a dataset of over one hundred thousand twitter messages collected during the U.S. State of the Union presidential address in 2010.

DimStiller is a system for dimensionality reduction and analysis. It frames the task of understanding and transforming input dimensions as a series of analysis steps where users transform data tables by chaining together different techniques, called operators, into pipelines of expressions. The individual operators have controls and views that are linked together based on the structure of the expression. Users interact with the operator controls to tune parameter choices, with immediate visual feedback guiding the exploration of local neighborhoods of the space of possible data tables. DimStiller also provides global guidance for navigating data-table space through expression templates called workflows, which permit re-use of common patterns of analysis.

Events that happened in the past are important for understanding the ongoing processes, predicting future developments, and making informed decisions. Significant and/or interesting events tend to attract many people. Some people leave traces of their attendance in the form of computer-processable data, such as records in the databases of mobile phone operators or photos on photo sharing web sites. We developed a suite of visual analytics methods for reconstructing past events from these activity traces. Our tools combine geocomputations, interactive geovisualizations and statistical methods to enable integrated analysis of the spatial, temporal, and thematic components of the data, including numeric attributes and texts. We demonstrate the utility of our approach on two large real data sets, mobile phone calls in Milano during 9 days and flickr photos made on British Isles during 5 years.

Although the discovery and analysis of communication patterns in large and complex email datasets are difficult tasks, they can be a valuable source of information. This paper presents EmailTime's capabilities through several examples. EmailTime is a visual analysis of email correspondence patterns over the course of time that interactively portrays personal and interpersonal networks using the correspondence in the email dataset. We suggest that integrating both statistics and visualizations in order to display information about the email datasets may simplify its evaluation.

We created a visual chat application for use during hazardous weather events. The application, NWSChat2, allows National Weather Service forecasters, media members, and storm trackers to communicate with each other, basing their conversation on a common shared radar map of the storm. Users can additionally annotate the map with `pins' or draw notes with a stylus. These annotations are automatically shared with all other users. The collaborative nature of NWSChat2 makes it well-suited for disseminating information to all users during weather emergencies.

This paper presents a case study with Enron email dataset to explore the behaviors of email users within different organizational positions. We defined email behavior as the email activity level of people regarding a series of measured metrics e.g. sent and received emails, numbers of email addresses, etc. These metrics were calculated through EmailTime, a visual analysis tool of email correspondence over the course of time. Results showed specific patterns in the email datasets of different organizational positions.

Data sets in astronomy are growing to enormous sizes. Modern astronomical surveys provide not only image data but also catalogues of millions of objects (stars, galaxies), each object with hundreds of associated parameters. Exploration of this very high-dimensional data space poses a huge challenge. Subspace clustering is one among several approaches which have been proposed for this purpose in recent years. However, many clustering algorithms require the user to set a large number of parameters without any guidelines. Some methods also do not provide a concise summary of the datasets, or, if they do, they lack additional important information such as the number of clusters present or the significance of the clusters. In this paper, we propose a method for ranking subspaces for clustering which overcomes many of the above limitations. First we carry out a transformation from parametric space to discrete image space where the data are represented by a grid-based density field. Then we apply so-called connected morphological operators on this density field of astronomical objects that provides visual support for the analysis of the important subspaces. Clusters in subspaces correspond to high-intensity regions in the density image. The importance of a cluster is measured by a new quality criterion based on the dynamics of local maxima of the density. Connected operators are able to extract such regions with an indication of the number of clusters present. The subspaces are visualized during computation of the quality measure, so that the user can interact with the system to improve the results. In the result stage, we use three visualization toolkits linked within a graphical user interface so that the user can perform an in-depth exploration of the ranked subspaces. Evaluation based on synthetic as well as real astronomical datasets demonstrates the power of the new method. We recover various known astronomical relations directly from the data with little or no a pri- - ori assumptions. Hence, our method holds good prospects for discovering new relations as well.

Visualization of multi-dimensional data is challenging due to the number of complex correlations that may be present in the data but that are difficult to be visually identified. One of the main causes for this problem is the inherent loss of information that occurs when high-dimensional data is projected into 2D or 3D. Although 2D scatterplots are ubiquitous due to their simplicity and familiarity, there are not a lot of variations on their basic metaphor. In this paper, we present a new way of visualizing multidimensional data using scatterplots. We extend 2D scatterplots using sensitivity coefficients to highlight local variation of one variable with respect to another. When applied to a scatterplot, these sensitivities can be understood as velocities, and the resulting visualization resembles a flow field. We also present a number of operations, based on flow-field analysis, that help users navigate, select and cluster points in an efficient manner. We show the flexibility and generality of this approach using a number of multidimensional data sets across different domains.

Information foraging and sensemaking with heterogeneous information are context-dependent activities. Thus visual analytics tools to support these activities must incorporate context. But, context is a difficult concept to define, model, and represent. Creating and representing context in support of visually-enabled reasoning about complex problems with complex information is a complementary but different challenge than that addressed in context-aware computing. In the latter, the goal is automated adaptation of the system to meet user needs for applications such as mobile location-based services where information about the location, the user, and the user goals filters what gets presented on a small mobile device. In contrast, for visual analytics-enabled information foraging and sensemaking, the user is likely to take an active role in foraging for the contextual information needed to support sensemaking in relation to some multifaceted problem. In this paper, we address the challenges of constructing and representing context within visual interfaces that support analytical reasoning in crisis management and humanitarian relief. The challenges stem from the diverse forms of information that can provide context and difficulty in defining and operationalizing context itself. Here, we pay particular attention to document foraging to support construction of the geographic and historical context within which monitoring and sensemaking can be carried out. Specifically, we present the concept of geo-historical context (GHC) and outline an empirical assessment of both the concept and its implementation in the Context Discovery Application, a web-based tool that supports document foraging and sensemaking.