IEEE VIS Publication Dataset

Text data such as online news and microblogs bear valuable insights regarding important events and responses to such events. Events are inherently temporal, evolving over time. Existing visual text analysis systems have provided temporal views of changes based on topical themes extracted from text data. But few have associated topical themes with events that cause the changes. In this paper, we propose an interactive visual analytics system, LeadLine, to automatically identify meaningful events in news and social media data and support exploration of the events. To characterize events, LeadLine integrates topic modeling, event detection, and named entity recognition techniques to automatically extract information regarding the investigative 4 Ws: who, what, when, and where for each event. To further support analysis of the text corpora through events, LeadLine allows users to interactively examine meaningful events using the 4 Ws to develop an understanding of how and why. Through representing large-scale text corpora in the form of meaningful events, LeadLine provides a concise summary of the corpora. LeadLine also supports the construction of simple narratives through the exploration of events. To demonstrate the efficacy of LeadLine in identifying events and supporting exploration, two case studies were conducted using news and social media data.

Extracting information from text is challenging. Most current practices treat text as a bag of words or word clusters, ignoring valuable linguistic information. Leveraging this linguistic information, we propose a novel approach to visualize textual information. The novelty lies in using state-of-the-art Natural Language Processing (NLP) tools to automatically annotate text which provides a basis for new and powerful interactive visualizations. Using NLP tools, we built a web-based interactive visual browser for human history articles from Wikipedia.

In recent years, the quantity of time series data generated in a wide variety of domains grown consistently. Thus, it is difficult for analysts to process and understand this overwhelming amount of data. In the specific case of time series data another problem arises: time series can be highly interrelated. This problem becomes even more challenging when a set of parameters influences the progression of a time series. However, while most visual analysis techniques support the analysis of short time periods, e.g. one day or one week, they fail to visualize large-scale time series, ranging over one year or more. In our approach we present a time series matrix visualization that tackles this problem. Its primary advantages are that it scales to a large number of time series with different start and end points and allows for the visual comparison / correlation analysis of a set of influencing factors. To evaluate our approach, we applied our technique to a real-world data set, showing the impact of local weather conditions on the efficiency of photovoltaic power plants.

Despite the extensive work done in the scientific visualization community on the creation and optimization of spatial data structures, there has been little adaptation of these structures in visual analytics and information visualization. In this work we present how we modify a space-partioning time (SPT) tree - a structure normally used in direct-volume rendering - for geospatial-temporal visualizations. We also present optimization techniques to improve the traversal speed of our structure through locational codes and bitwise comparisons. Finally, we present the results of an experiment that quantitatively evaluates our modified SPT tree with and without our optimizations. Our results indicate that retrieval was nearly three times faster when using our optimizations, and are consistent across multiple trials. Our finding could have implications for performance in using our modified SPT tree in large-scale geospatial temporal visual analytics software.

Recent research suggests that the personality trait Locus of Control (LOC) can be a reliable predictor of performance when learning a new visualization tool. While these results are compelling and have direct implications to visualization design, the relationship between a user's LOC measure and their performance is not well understood. We hypothesize that there is a dependent relationship between LOC and performance; specifically, a person's orientation on the LOC scale directly influences their performance when learning new visualizations. To test this hypothesis, we conduct an experiment with 300 subjects using Amazon's Mechanical Turk. We adapt techniques from personality psychology to manipulate a user's LOC so that users are either primed to be more internally or externally oriented on the LOC scale. Replicating previous studies investigating the effect of LOC on performance, we measure users' speed and accuracy as they use visualizations with varying visual metaphors. Our findings demonstrate that changing a user's LOC impacts their performance. We find that a change in users' LOC results in performance changes.

Contingency tables summarize the relations between categorical variables and arise in both scientific and business domains. Asymmetrically large two-way contingency tables pose a problem for common visualization methods. The Contingency Wheel has been recently proposed as an interactive visual method to explore and analyze such tables. However, the scalability and readability of this method are limited when dealing with large and dense tables. In this paper we present Contingency Wheel++, new visual analytics methods that overcome these major shortcomings: (1) regarding automated methods, a measure of association based on Pearson's residuals alleviates the bias of the raw residuals originally used, (2) regarding visualization methods, a frequency-based abstraction of the visual elements eliminates overlapping and makes analyzing both positive and negative associations possible, and (3) regarding the interactive exploration environment, a multi-level overview+detail interface enables exploring individual data items that are aggregated in the visualization or in the table using coordinated views. We illustrate the applicability of these new methods with a use case and show how they enable discovering and analyzing nontrivial patterns and associations in large categorical data.

The Common N-Gram (CNG) classifier is a text classification algorithm based on the comparison of frequencies of character n-grams (strings of characters of length n) that are the most common in the considered documents and classes of documents. We present a text analytic visualization system that employs the CNG approach for text classification and uses the differences in frequency values of common n-grams in order to visually compare documents at the sub-word level. The visualization method provides both an insight into n-gram characteristics of documents or classes of documents and a visual interpretation of the workings of the CNG classifier.

Significant effort has been devoted to designing clustering algorithms that are responsive to user feedback or that incorporate prior domain knowledge in the form of constraints. However, users desire more expressive forms of interaction to influence clustering outcomes. In our experiences working with diverse application scientists, we have identified an interaction style scatter/gather clustering that helps users iteratively restructure clustering results to meet their expectations. As the names indicate, scatter and gather are dual primitives that describe whether clusters in a current segmentation should be broken up further or, alternatively, brought back together. By combining scatter and gather operations in a single step, we support very expressive dynamic restructurings of data. Scatter/gather clustering is implemented using a nonlinear optimization framework that achieves both locality of clusters and satisfaction of user-supplied constraints. We illustrate the use of our scatter/gather clustering approach in a visual analytic application to study baffle shapes in the bat biosonar (ears and nose) system. We demonstrate how domain experts are adept at supplying scatter/gather constraints, and how our framework incorporates these constraints effectively without requiring numerous instance-level constraints.

Visual analytic tools aim to support the cognitively demanding task of sensemaking. Their success often depends on the ability to leverage capabilities of mathematical models, visualization, and human intuition through flexible, usable, and expressive interactions. Spatially clustering data is one effective metaphor for users to explore similarity and relationships between information, adjusting the weighting of dimensions or characteristics of the dataset to observe the change in the spatial layout. Semantic interaction is an approach to user interaction in such spatializations that couples these parametric modifications of the clustering model with users' analytic operations on the data (e.g., direct document movement in the spatialization, highlighting text, search, etc.). In this paper, we present results of a user study exploring the ability of semantic interaction in a visual analytic prototype, ForceSPIRE, to support sensemaking. We found that semantic interaction captures the analytical reasoning of the user through keyword weighting, and aids the user in co-creating a spatialization based on the user's reasoning and intuition.

Due to the ever growing volume of acquired data and information, users have to be constantly aware of the methods for their exploration and for interaction. Of these, not each might be applicable to the data at hand or might reveal the desired result. Owing to this, innovations may be used inappropriately and users may become skeptical. In this paper we propose a knowledge-assisted interface for medical visualization, which reduces the necessary effort to use new visualization methods, by providing only the most relevant ones in a smart way. Consequently, we are able to expand such a system with innovations without the users to worry about when, where, and especially how they may or should use them. We present an application of our system in the medical domain and give qualitative feedback from domain experts.

Increasingly, social network datasets contain social attribute information about actors and their relationship. Analyzing such network with social attributes requires making sense of not only its structural features, but also the relationship between social features in attributes and network structures. Existing social network analysis tools are usually weak in supporting complex analytical tasks involving both structural and social features, and often overlook users' needs for sensemaking tools that help to gather, synthesize, and organize information of these features. To address these challenges, we propose a sensemaking framework of social-network visual analytics in this paper. This framework considers both bottom-up processes, which are about constructing new understandings based on collected information, and top-down processes, which concern using prior knowledge to guide information collection, in analyzing social networks from both social and structural perspectives. The framework also emphasizes the externalization of sensemaking processes through interactive visualization. Guided by the framework, we develop a system, SocialNetSense, to support the sensemaking in visual analytics of social networks with social attributes. The example of using our system to analyze a scholar collaboration network shows that our approach can help users gain insight into social networks both structurally and socially, and enhance their process awareness in visual analytics.

Recent advances in technology have enabled social media services to support space-time indexed data, and internet users from all over the world have created a large volume of time-stamped, geo-located data. Such spatiotemporal data has immense value for increasing situational awareness of local events, providing insights for investigations and understanding the extent of incidents, their severity, and consequences, as well as their time-evolving nature. In analyzing social media data, researchers have mainly focused on finding temporal trends according to volume-based importance. Hence, a relatively small volume of relevant messages may easily be obscured by a huge data set indicating normal situations. In this paper, we present a visual analytics approach that provides users with scalable and interactive social media data analysis and visualization including the exploration and examination of abnormal topics and events within various social media data sources, such as Twitter, Flickr and YouTube. In order to find and understand abnormal events, the analyst can first extract major topics from a set of selected messages and rank them probabilistically using Latent Dirichlet Allocation. He can then apply seasonal trend decomposition together with traditional control chart methods to find unusual peaks and outliers within topic time series. Our case studies show that situational awareness can be improved by incorporating the anomaly and trend examination techniques into a highly interactive visual analysis process.

In explorative data analysis, the data under consideration often resides in a high-dimensional (HD) data space. Currently many methods are available to analyze this type of data. So far, proposed automatic approaches include dimensionality reduction and cluster analysis, whereby visual-interactive methods aim to provide effective visual mappings to show, relate, and navigate HD data. Furthermore, almost all of these methods conduct the analysis from a singular perspective, meaning that they consider the data in either the original HD data space, or a reduced version thereof. Additionally, HD data spaces often consist of combined features that measure different properties, in which case the particular relationships between the various properties may not be clear to the analysts a priori since it can only be revealed if appropriate feature combinations (subspaces) of the data are taken into consideration. Considering just a single subspace is, however, often not sufficient since different subspaces may show complementary, conjointly, or contradicting relations between data items. Useful information may consequently remain embedded in sets of subspaces of a given HD input data space. Relying on the notion of subspaces, we propose a novel method for the visual analysis of HD data in which we employ an interestingness-guided subspace search algorithm to detect a candidate set of subspaces. Based on appropriately defined subspace similarity functions, we visualize the subspaces and provide navigation facilities to interactively explore large sets of subspaces. Our approach allows users to effectively compare and relate subspaces with respect to involved dimensions and clusters of objects. We apply our approach to synthetic and real data sets. We thereby demonstrate its support for understanding HD data from different perspectives, effectively yielding a more complete view on HD data.

Reconstruction of shredded documents remains a significant challenge. Creating a better document reconstruction system enables not just recovery of information accidentally lost but also understanding our limitations against adversaries' attempts to gain access to information. Existing approaches to reconstructing shredded documents adopt either a predominantly manual (e.g., crowd-sourcing) or a near automatic approach. We describe Deshredder, a visual analytic approach that scales well and effectively incorporates user input to direct the reconstruction process. Deshredder represents shredded pieces as time series and uses nearest neighbor matching techniques that enable matching both the contours of shredded pieces as well as the content of shreds themselves. More importantly, Deshred-der's interface support visual analytics through user interaction with similarity matrices as well as higher level assembly through more complex stitching functions. We identify a functional task taxonomy leading to design considerations for constructing deshredding solutions, and describe how Deshredder applies to problems from the DARPA Shredder Challenge through expert evaluations.

Event data can hold valuable decision making information, yet detecting interesting patterns in this type of data is not an easy task because the data is usually rich and contains spatial, temporal as well as multivariate dimensions. Research into visual analytics tools to support the discovery of patterns in event data often focuses on the spatiotemporal or spatiomultivariate dimension of the data only. Few research efforts focus on all three dimensions in one framework. An integral view on all three dimensions is, however, required to unlock the full potential of event datasets. In this poster, we present an event visualization, transition, and interaction framework that enables an integral view on all dimensions of spatiotemporal multivariate event data. The framework is built around the notion that the event data space can be considered a spatiotemporal multivariate hypercube. Results of a case study we performed suggest that a visual analytics tool based on the proposed framework is indeed capable to support users in the discovery of multidimensional spatiotemporal multivariate patterns in event data.

Visual analytics emphasizes the interplay between visualization, analytical procedures performed by computers and human perceptual and cognitive activities. Human reasoning is an important element in this context. There are several theories in psychology and HCI explaining open-ended and exploratory reasoning. Five of these theories (sensemaking theories, gestalt theories, distributed cognition, graph comprehension theories and skill-rule-knowledge models) are described in this paper. We discuss their relevance for visual analytics. In order to do this more systematically, we developed a schema of categories relevant for visual analytics research and evaluation. All these theories have strengths but also weaknesses in explaining interaction with visual analytics systems. A possibility to overcome the weaknesses would be to combine two or more of these theories.

Temporal awareness is pivotal to successful real-time dynamic decision making in a wide range of command and control situations; particularly in safety-critical environments. However, little explicit support for operators' temporal awareness is provided by decision support systems (DSS) for time-critical decisions. In the context of functional simulations of naval anti-air warfare and emergency response management, the present study compares operator support provided by two display formats. In both environments, we contrast a baseline condition to a condition in which a temporal display was integrated to the original interface to support operators' temporal awareness. We also wish to establish whether the implementation of time-based DSSs may also come with drawbacks on cognitive functioning and performance.

We introduce translational science, a research discipline from medicine, and show how adapting it for visual analytics can improve the design and evaluation of visual analytics interfaces. Translational science “translates” knowledge from the lab to the real-world to “ground truth” by incorporating a 3 phase program of research. Phase 1 & 2 include protocols for research in the lab and field and Phase 3 focuses on dissemination and documentation. We discuss these phases and how they may be applied to visual analytics research.

Datasets that are collected for research often contain millions of records and may carry hidden pitfalls that are hard to detect. This work demonstrates how visual analytics can be used for identifying problems in the spatial distribution of crawled photographic data in different datasets: Picasa Web Albums, Panoramio, Flickr and Geograph, chosen to be potential data sources for ongoing doctoral research. This poster summary describes a number of problems found in the datasets using visual analytics and suggests that greater attention should be paid to assessing the quality of data gathered from user-generated photographic content. This work is the first part of a three-year PhD project aimed at producing a pedestrian-routing system that can suggest attractive pathways extracted from user-generated photographic content.

Data quality evaluation is one of the most critical steps during the data mining processes. Data with poor quality often leads to poor performance in data mining, low efficiency in data analysis, wrong decision which bring great economic loss to users and organizations further. Although many researches have been carried out from various aspects of the extracting, transforming, and loading processes in data mining, most researches pay more attention to analysis automation than to data quality evaluation. To address the data quality evaluation issues, we propose an approach to combine human beings' powerful cognitive abilities in data quality evaluation with the high efficiency ability of computer, and develop a visual analysis method for data quality evaluation based on visual morphology.