IEEE VIS Publication Dataset

An algorithm is presented for the visualisation of multidimensional abstract data, building on a hybrid model introduced at InfoVis 2002. The most computationally complex stage of the original model involved performing a nearest-neighbour search for every data item. The complexity of this phase has been reduced by treating all high-dimensional relationships as a set of discretised distances to a constant number of randomly selected pivot items. In improving this computational bottleneck, the complexity is reduced from O(N v N) to O(N 5 4 ). As well as documenting this improvement, the paper describes evaluation with a data set of 108000 14-dimensional items; a considerable increase on the size of data previously tested. Results illustrate that the reduction in complexity is reflected in significantly improved run times and that no negative impact is made upon the quality of layout produced

Large and high-dimensional data sets mapped to low-dimensional visualizations often result in perceptual ambiguities. One such ambiguity is overlap or occlusion that occurs when the number of records exceeds the number of unique locations in the presentation or when there exist two or more records that map to the same location. To lessen the affect of occlusion, non-standard visual attributes (i.e. shading and/or transparency) are applied, or such records may be remapped to a corresponding jittered location. The resulting mapping efficiently portrays the crowding of records but fails to provide the insight into the relationship between the neighboring records. We introduce a new interactive technique that intelligibly organizes overlapped points, a neural network-based smart jittering algorithm. We demonstrate this technique on a scatter plot, the most widely used visualization. The algorithm can be applied to other one, two, and multi-dimensional visualizations which represent data as points, including 3-dimensional scatter plots, RadViz, polar coordinates.

Large number of dimensions not only cause clutter in multi-dimensional visualizations, but also make it difficult for users to navigate the data space. Effective dimension management, such as dimension ordering, spacing and filtering, is critical for visual exploration of such datasets. Dimension ordering and spacing explicitly reveal dimension relationships in arrangement-sensitive multidimensional visualization techniques, such as parallel coordinates, star glyphs, and pixel-oriented techniques. They facilitate the visual discovery of patterns within the data. Dimension filtering hides some of the dimensions to reduce clutter while preserving the major information of the dataset. In this paper, we propose an interactive hierarchical dimension ordering, spacing and filtering approach, called DOSFA. DOSFA is based on dimension hierarchies derived from similarities among dimensions. It is scalable multi-resolution approach making dimensional management a tractable task. On the one hand, it automatically generates default settings for dimension ordering, spacing and filtering. On the other hand, it allows users to efficiently control all aspects of this dimension management process via visual interaction tools for dimension hierarchy manipulation. A case study visualizing a dataset containing over 200 dimensions reveals high dimensional visualization techniques.

Data sets with a large number of nominal variables, some with high cardinality, are becoming increasingly common and need to be explored. Unfortunately, most existing visual exploration displays are designed to handle numeric variables only. When importing data sets with nominal values into such visualization tools, most solutions to date are rather simplistic. Often, techniques that map nominal values to numbers do not assign order or spacing among the values in a manner that conveys semantic relationships. Moreover, displays designed for nominal variables usually cannot handle high cardinality variables well. This paper addresses the problem of how to display nominal variables in general-purpose visual exploration tools designed for numeric variables. Specifically, we investigate (1) how to assign order and spacing among the nominal values, and (2) how to reduce the number of distinct values to display. We propose that nominal variables be pre-processed using a distance-quantification-classing (DQC) approach before being imported into a visual exploration tool. In the distance step, we identify a set of independent dimensions that can be used to calculate the distance between nominal values. In the quantification step, we use the independent dimensions and the distance information to assign order and spacing among the nominal values. In the classing step, we use results from the previous steps to determine which values within a variable are similar to each other and thus can be grouped together. Each step in the DQC approach can be accomplished by a variety of techniques. We extended the XmdvTool package to incorporate this approach. We evaluated our approach on several data sets using a variety of evaluation measures.

Graph and tree visualization techniques enable interactive exploration of complex relations while communicating topology. However, most existing techniques have not been designed for situations where visual information such as images is also present at each node and must be displayed. This paper presents MoireGraphs to address this need. MoireGraphs combine a new focus+context radial graph layout with a suite of interaction techniques (focus strength changing, radial rotation, level highlighting, secondary foci, animated transitions and node information) to assist in the exploration of graphs with visual nodes. The method is scalable to hundreds of displayed visual nodes.

Many networks under study in Information Visualization are "small world" networks. These networks first appeared in the study social networks and were shown to be relevant models in other application domains such as software reverse engineering and biology. Furthermore, many of these networks actually have a multiscale nature: they can be viewed as a network of groups that are themselves small world networks. We describe a metric that has been designed in order to identify the weakest edges in a small world network leading to an easy and low cost filtering procedure that breaks up a graph into smaller and highly connected components. We show how this metric can be exploited through an interactive navigation of the network based on semantic zooming. Once the network is decomposed into a hierarchy of sub-networks, a user can easily find groups and subgroups of actors and understand their dynamics.

Large 2D information spaces, such as maps, images, or abstract visualizations, require views at various level of detail: close ups to inspect details, overviews to maintain (literally) an overview. Users often switch between these views. We discuss how smooth animations from one view to another can be defined. To this end, a metric on the effect of simultaneous zooming and panning is defined, based on an estimate of the perceived velocity. Optimal is defined as smooth and efficient. Given the metric, these terms can be translated into a computational model, which is used to calculate an analytic solution for optimal animations. The model has two free parameters: animation speed and zoom/pan trade off. A user experiment to find good values for these is described.

This paper describes Thread Arcs, a novel interactive visualization technique designed to help people use threads found in email. Thread Arcs combine the chronology of messages with the branching tree structure of a conversational thread in a mixed-model visualization by Venolia and Neustaedter (2003) that is stable and compact. By quickly scanning and interacting with Thread Arcs, people can see various attributes of conversations and find relevant messages in them easily. We tested this technique against other visualization techniques with users' own email in a functional prototype email client. Thread Arcs proved an excellent match for the types of threads found in users' email for the qualities users wanted in small-scale visualizations.

Traditionally, node link diagrams are the prime choice when it comes to visualizing software architectures. However, node link diagrams often fall short when used to visualize large graph structures. In this paper we investigate the use of call matrices as visual aids in the management of large software projects. We argue that call matrices have a number of advantages over traditional node link diagrams when the main object of interest is the link instead of the node. Matrix visualizations can provide stable and crisp layouts of large graphs and are inherently well suited for large multilevel visualizations because of their recursive structure. We discuss a number of visualization issues, using a very large software project currently under development at Philips Medical Systems as a running example.

We present a novel family of data-driven linear transformations, aimed at visualizing multivariate data in a low-dimensional space in a way that optimally preserves the structure of the data. The well-studied PCA and Fisher's LDA are shown to be special members in this family of transformations, and we demonstrate how to generalize these two methods such as to enhance their performance. Furthermore, our technique is the only one, to the best of our knowledge, that reflects in the resulting embedding both the data coordi-nates and pairwise similarities and/or dissimilarities between the data elements. Even more so, when information on the clustering (labeling) decomposition of the data is known, this information can be integrated in the linear transformation, resulting in embeddings that clearly show the separation between the clusters, as well as their intra-structure. All this makes our technique very flexible and powerful, and lets us cope with kinds of data that other techniques fail to describe properly.

Satisfaction surveys are an important measurement tool in fields such as market research or human resources management. Serious studies consist of numerous questions and contain answers from large population samples. Aggregation on both sides, the questions asked as well as the answers received, turns the multidimensional problem into a complex system of interleaved hierarchies. Traditional ways of presenting the results are limited to one-dimensional charts and cross-tables. We developed a visualization method called the Parallel Coordinate Tree that combines multidimensional analysis with a tree structure representation. Distortion-oriented focus+context techniques are used to facilitate interaction with the visualization. In this paper we present a design study of a commercial application that we built, using this method to analyze and communicate results from large-scale customer satisfaction surveys.

Network evolution is an ubiquitous phenomenon in a wide variety of complex systems. There is an increasing interest in statistically modeling the evolution of complex networks such as small-world networks and scale-free networks. In this article, we address a practical issue concerning the visualizations of co-citation networks of scientific publications derived by two widely known link reduction algorithms, namely minimum spanning trees (MSTs) and pathfinder networks (PFNETs). Our primary goal is to identify the strengths and weaknesses of the two methods in fulfilling the need for visualizing evolving networks. Two criteria are derived for assessing visualizations of evolving networks in terms of topological properties and dynamical properties. We examine the animated visualization models of the evolution of botulinum toxin research in terms of its co-citation structure across a 58-year span (1945-2002). The results suggest that although high-degree nodes dominate the structure of MST models, such structures can be inadequate in depicting the essence of how the network evolves because MST removes potentially significant links from high-order shortest paths. In contrast, PFNET models clearly demonstrate their superiority in maintaining the cohesiveness of some of the most pivotal paths, which in turn make the growth animation more predictable and interpretable. We suggest that the design of visualization and modeling tools for network evolution should take the cohesiveness of critical paths into account.

We present a hardware-accelerated method for visualizing 3D flow fields. The method is based on insertion, advection, and decay of dye. To this aim, we extend the texture-based IBFV technique presented by van Wijk (2001) for 2D flow visualization in two main directions. First, we decompose the 3D flow visualization problem in a series of 2D instances of the mentioned IBFV technique. This makes our method benefit from the hardware acceleration the original IBFV technique introduced. Secondly, we extend the concept of advected gray value (or color) noise by introducing opacity (or matter) noise. This allows us to produce sparse 3D noise pattern advections, thus address the occlusion problem inherent to 3D flow visualization. Overall, the presented method delivers interactively animated 3D flow, uses only standard OpenGL 1.1 calls and 2D textures, and is simple to understand and implement.

We present a haptic rendering technique that uses directional constraints to facilitate enhanced exploration modes for volumetric datasets. The algorithm restricts user motion in certain directions by incrementally moving a proxy point along the axes of a local reference frame. Reaction forces are generated by a spring coupler between the proxy and the data probe, which can be tuned to the capabilities of the haptic interface. Secondary haptic effects including field forces, friction, and texture can be easily incorporated to convey information about additional characteristics of the data. We illustrate the technique with two examples: displaying fiber orientation in heart muscle layers and exploring diffusion tensor fiber tracts in brain white matter tissue. Initial evaluation of the approach indicates that haptic constraints provide an intuitive means or displaying directional information in volume data.

We present an innovative modeling and rendering primitive, called the O-buffer, for sample-based graphics, such as images, volumes and points. The 2D or 3D O-buffer is in essence a conventional image or a volume, respectively, except that samples are not restricted to a regular grid. A sample position in the O-buffer is recorded as an offset to the nearest grid point of a regular base grid (hence the name O-buffer). The offset is typically quantized for compact representation and efficient rendering. The O-buffer emancipates pixels and voxels from the regular grids and can greatly improve the modeling power of images and volumes. It is a semi-regular structure which lends itself to efficient construction and rendering. Image quality can be improved by storing more spatial information with samples and by avoiding multiple resamplings and delaying reconstruction to the final rendering stage. Using O-buffers, more accurate multi-resolution representations can be developed for images and volumes. It can also be exploited to represent and render unstructured primitives, such as points, particles, curvilinear or irregular volumes. The O-buffer is therefore a uniform representation for a variety of graphics primitives and supports mixing them in the same scene. We demonstrate the effectiveness of the O-buffer with hierarchical O-buffers, layered depth O-buffers, and hybrid volume rendering with O-buffers.

This paper introduces a method for converting an image or volume sampled on a regular grid into a space-efficient irregular point hierarchy. The conversion process retains the original frequency characteristics of the dataset by matching the spatial distribution of sample points with the required frequency. To achieve good blending, the spherical points commonly used in volume rendering are generalized to ellipsoidal point primitives. A family of multiresolution, oriented Gabor wavelets provide the frequency-space analysis of the dataset. The outcome of this frequency analysis is the reduced set of points, in which the sampling rate is decreased in originally oversampled areas. During rendering, the traversal of the hierarchy can be controlled by any suitable error metric or quality criteria. The local level of refinement is also sensitive to the transfer function. Areas with density ranges mapped to high transfer function variability are rendered at higher point resolution than others. Our decomposition is flexible and can be used for iso-surface rendering, alpha compositing and X-ray rendering of volumes. We demonstrate our hierarchy with an interactive splatting volume renderer, in which the traversal of the point hierarchy for rendering is modulated by a user-specified frame rate.

We combine topological and geometric methods to construct a multi-resolution data structure for functions over two-dimensional domains. Starting with the Morse-Smale complex, we construct a topological hierarchy by progressively canceling critical points in pairs. Concurrently, we create a geometric hierarchy by adapting the geometry to the changes in topology. The data structure supports mesh traversal operations similarly to traditional multi-resolution representations.

In the traditional volume visualization paradigm, the user specifies a transfer function that assigns each scalar value to a color and opacity by defining an opacity and a color map function. The transfer function has two limitations. First, the user must define curves based on histogram and value rather than seeing and working with the volume itself. Second, the transfer function is inflexible in classifying regions of interest, where values at a voxel such as intensity and gradient are used to differentiate material, not talking into account additional properties such as texture and position. We describe an intuitive user interface for specifying the classification functions that consists of the users painting directly on sample slices of the volume. These painted regions are used to automatically define high-dimensional classification functions that can be implemented in hardware for interactive rendering. The classification of the volume is iteratively improved as the user paints samples, allowing intuitive and efficient viewing of materials of interest.