Appendectomy is the most common abdominal surgical procedure performed in children in the United States. In order to assist care providers in creating treatment plans for the postoperative management of pediatric appendicitis, we have developed a predictive statistical model of outcomes on which we have built a prototype decision aid application. The model, trained on 3724 anonymized care records and evaluated on a separate set of 2205 cases from a tertiary care center, achieves 97.0% specificity, 25.1% true sensitivity, and 58.8% precision. We have also built an interactive decision support tool augmented with simple visualization techniques designed for clinicians to use in the course of making care decisions (e.g., discharge) and in patient/stakeholder communication. Its focus is on end-user ease of use and integration into existing clinician workflows, and is designed to evolve its predictions as more and better data become available.

In this article, we present an interactive visual information retrieval and recommendation system, called VisIRR, for large-scale document discovery. VisIRR effectively combines the paradigms of (1) a passive pull through query processes for retrieval and (2) an active push that recommends items of potential interest to users based on their preferences. Equipped with an efficient dynamic query interface against a large-scale corpus, VisIRR organizes the retrieved documents into high-level topics and visualizes them in a 2D space, representing the relationships among the topics along with their keyword summary. In addition, based on interactive personalized preference feedback with regard to documents, VisIRR provides document recommendations from the entire corpus, which are beyond the retrieved sets. Such recommended documents are visualized in the same space as the retrieved documents, so that users can seamlessly analyze both existing and newly recommended ones. This article presents novel computational methods, which make these integrated representations and fast interactions possible for a large-scale document corpus. We illustrate how the system works by providing detailed usage scenarios. Additionally, we present preliminary user study results for evaluating the effectiveness of the system.

ObjectiveThe objective of this analysis is to leverage recent advances innatural language processing (NLP) to develop new methods andsystem capabilities for processing social media (Twitter messages)for situational awareness (SA), syndromic surveillance (SS), andevent-based surveillance (EBS). Specifically, we evaluated the useof human-in-the-loop semantic analysis to assist public health (PH)SA stakeholders in SS and EBS using massive amounts of publiclyavailable social media data.IntroductionSocial media messages are often short, informal, and ungrammatical.They frequently involve text, images, audio, or video, which makesthe identification of useful information difficult. This complexityreduces the efficacy of standard information extraction techniques1.However, recent advances in NLP, especially methods tailoredto social media2, have shown promise in improving real-time PHsurveillance and emergency response3. Surveillance data derived fromsemantic analysis combined with traditional surveillance processeshas potential to improve event detection and characterization. TheCDC Office of Public Health Preparedness and Response (OPHPR),Division of Emergency Operations (DEO) and the Georgia TechResearch Institute have collaborated on the advancement of PH SAthrough development of new approaches in using semantic analysisfor social media.MethodsTo understand how computational methods may benefit SS andEBS, we studied an iterative refinement process, in which the datauser actively cultivated text-based topics (“semantic culling”) in asemi-automated SS process. This ‘human-in-the-loop’ process wascritical for creating accurate and efficient extraction functions in large,dynamic volumes of data. The general process involved identifyinga set of expert-supplied keywords, which were used to collect aninitial set of social media messages. For purposes of this analysisresearchers applied topic modeling to categorize related messages intoclusters. Topic modeling uses statistical techniques to semanticallycluster and automatically determine salient aggregations. A user thensemantically culled messages according to their PH relevance.In June 2016, researchers collected 7,489 worldwide English-language Twitter messages (tweets) and compared three samplingmethods: a baseline random sample (C1, n=2700), a keyword-basedsample (C2, n=2689), and one gathered after semantically cullingC2 topics of irrelevant messages (C3, n=2100). Researchers utilizeda software tool, Luminoso Compass4, to sample and perform topicmodeling using its real-time modeling and Twitter integrationfeatures. For C2 and C3, researchers sampled tweets that theLuminoso service matched to both clinical and layman definitions ofRash, Gastro-Intestinal syndromes5, and Zika-like symptoms. Laymanterms were derived from clinical definitions from plain languagemedical thesauri. ANOVA statistics were calculated using SPSSsoftware, version. Post-hoc pairwise comparisons were completedusing ANOVA Turkey’s honest significant difference (HSD) test.ResultsAn ANOVA was conducted, finding the following mean relevancevalues: 3% (+/- 0.01%), 24% (+/- 6.6%) and 27% (+/- 9.4%)respectively for C1, C2, and C3. Post-hoc pairwise comparison testsshowed the percentages of discovered messages related to the eventtweets using C2 and C3 methods were significantly higher than forthe C1 method (random sampling) (p<0.05). This indicates that thehuman-in-the-loop approach provides benefits in filtering socialmedia data for SS and ESB; notably, this increase is on the basis ofa single iteration of semantic culling; subsequent iterations could beexpected to increase the benefits.ConclusionsThis work demonstrates the benefits of incorporating non-traditional data sources into SS and EBS. It was shown that an NLP-based extraction method in combination with human-in-the-loopsemantic analysis may enhance the potential value of social media(Twitter) for SS and EBS. It also supports the claim that advancedanalytical tools for processing non-traditional SA, SS, and EBSsources, including social media, have the potential to enhance diseasedetection, risk assessment, and decision support, by reducing the timeit takes to identify public health events.

We present the Test Matrix Tool (TMT) framework, a simulation-agnostic framework providing end-to-end support for robust analysis of complex systems. The need to execute a large number of simulations is common to many problem environments, even those already reduced by Design of Experiments or similar methodologies. TMT addresses key end-user needs in easing the specification, execution and analysis of simulation workloads in ways that are consistent between specific applications of the framework. The TMT design contributes modular specifications for key data communicated between and within the specification, execution and analysis components. Our TMT implementation is an instantiation of those formats freely available for general use. TMT's data analysis component provides a variety of features data filtering, comparison, transformation and visualization for analytic tasks on any TMT-embedded model. We provide a brief case study as an example of its use in a real-world application.

We present Lytic, a domain-independent, faceted visual analytic (VA) system for interactive exploration of large datasets. It combines a flexible UI that adapts to arbitrary character-separated value (CSV) datasets with algorithmic preprocessing to compute unsupervised dimension reduction and cluster data from high-dimensional fields. It provides a variety of visualization options that require minimal user effort to configure and a consistent user experience between visualization types and underlying datasets. Filtering, comparison and visualization operations work in concert, allowing users to hop seamlessly between actions and pursue answers to expected and unexpected data hypotheses.

Research areas: Machine learning, Data mining, Information visualization, Visual analytics, Text visualization.

Faceted metadata and navigation have become major topics in library science, information retrieval and Human-Computer Interaction (HCI). This work surveys a range of extant approaches in this design space, classifying systems along several relevant dimensions. We use that survey to analyze the organization of data and its querying within faceted browsing systems. We contribute formal entity-relationship (ER) and relational data models that explain that organization and relational query models that explain systems' browsing functionality. We use these types of models since they are widely used to conceptualize data and to model back-end data stores. Their structured nature also suggests ways in which both the models and faceted systems might be extended.

Hierarchical representations are common in digital repositories, yet are not always fully leveraged in their online search interfaces. This work describes ResultMaps, which use hierarchical treemap representations with query string-driven digital library search engines. We describe two lab experiments, which find that ResultsMap users yield significantly better results over a control condition on some subjective measures, and we find evidence that ResultMaps have ancillary benefits via increased understanding of some aspects of repository content. The ResultMap system and experiments contribute an understanding of the benefits-direct and indirect-of the ResultMap approach to repository search visualization.

This work presents our research into visualization for digital repository search interfaces, motivated by the prevalence of existing hierarchical data structures and the general lack of contextualization present in existing systems. We develop the ResultMap concept, a treemap-based visualization system that we have applied to keyword search engine and faceted classification data environments, and present the results of their empirical evaluation. We organize this work as follows: Chapter 1 provides an introduction to our problem area, motivates our general approach of leveraging hierarchical structure (via ResultMaps) for context, and proposes a thesis statement and corresponding research questions. Chapter 2 discusses related work, and includes a survey and design characterization of faceted navigation tools. Chapter 3 defines the key visual and interactive features of the ResultMap concept and justifies their basic design. Chapter 4 presents our implementation and evaluation of ResultMaps applied to digital library search engine result pages (SERPs). Chapter 5 consists of two major portions: a presentation of our formal data and query models for faceted environments and our implementation and evaluation of ResultMaps in a faceted UI context. In Chapter 6 we conclude—based on our results from Chapter 4 and Chapter 5—with a set of principles for designing both visual search interfaces themselves and designing their evaluation. We finish with suggestions for future research in this area.

ResultMaps are a treemap-based [5] search visualization concept for online digital repositories, designed to leverages hierarchical metadata often already present in digital libraries. We have previously reported our application of ResultMaps to a digital library’s keyword search engine result pages (SERPs) [3]. Here we discuss the use of ResultMaps in the more interesting context of faceted metadata and plans for their formal evaluation. ResultMaps perform several useful functions in that environment: placing items within a consistent visual context; provide a mechanism for outlier, cluster and correlation detection; and act as a visual medium for previewing prospective selections.

Digital libraries commonly elide hierarchical metadata that might be used more effectively. This proposal presents the ResultMap concept, a tool that leverages that metadata for digital library search facilities; an initial study of its effectiveness; the concept of applying ResultMaps to faceted metadata, allowing visual detection of implicit correlations between facets; and proposals for further study of ResultMaps in both directed search and faceted browsing environments. CR

Though attention to evaluating human-robot interfaces has increased in recent years, there are relatively few reports of using evaluation tools during the development of human-robot interaction (HRI) systems to gauge and improve their designs— possibly due to a shortage of suitable evaluation techniques. Heuristic evaluation is a technique suitable for such applications that has become popular in the human-computer interaction (HCI) community. However, it requires usability heuristics applicable to the system environment. This work contributes a set of heuristics appropriate for use with HRI systems, derived from a variety of sources both in and out of the HRI field. Evaluators have successfully used the heuristics on an HRI system, demonstrating their effectiveness against standard measures of heuristic effectiveness.

MacKenzie and Soukoreff have previously introduced a Fitts' Law-based performance model of expert two-thumb text entry on mini-QWERTY keyboards [4]. In this work we validate the original model using results from a longitudinal study of mini-QWERTY keyboards, and update the model to account for observed inter-key time data.

The input capabilities of mobile phones are limited by their physical form factor. Approaches to augmenting those capabilities that expand the input space without negatively impacting size or weight are particularly desirable. We propose adding simple pressure sensors under the keypad buttons to provide multiple channels of continuous pressure input. Pressure input supports a larger and more interesting interaction space without some of the unusual or unwanted qualities of some other approaches. We describe an implementation of our pressure-augmented system and show a number of interaction techniques, some old and some new, facilitated by continual pressure. We contrast these techniques with previous sensor-augmentation devices and highlight notable differences and advantages. ACM Classification: H5.2 [Information interfaces and presentation]: User Interfaces – Interaction styles, Input devices and strategies.

Browsing is a widespread user behavior in the digital library (DL) environment; there are an array of existing techniques that afford browsing and are readily applicable to digital libraries. We outline the designs of two such methods based on well-known techniques: treemaps and ScentTrails.

Digital libraries have great potential to improve the educational experience. As a result, there are a wide variety of such repositories, especially those that focus specifically on education. But relatively few focus on topics as specific as Human-Computer Interaction (HCI) or Human-Centered Computing (HCC). In addition, support for browsing behavior, with a few exceptions, is both weak and not suitable for user needs. This paper presents our work to create a repository of educational materials for a relatively narrowly-targeted field (HCC/HCI), including our requirements gathering methods and results. Finally, we discuss the HCC Education Digital Library (HCC EDL) as a platform for investigating broader digital library research questions, such as exploring alternative designs for content browsing mechanisms.

INTRODUCTION Technology has the potential to improve both the efficacy and efficiency of the educational experience from the perspective of both students and teachers. Specifically, educational repositories can enhance the quality of education through the provision of a diverse set of learning materials, and enhance the efficiency of education by affording the reuse of stored objects. To that end, we have created the Georgia Tech Human-Centered Computing (HCC) Education Digital Library (HCC EDL) [12] for use by the worldwide HCC and Human-Computer Interaction (HCI) communities.

We present a longitudinal study of mini-QWERTY keyboard use, examining the learning rates of novice mini-QWERTY users. The study consists of 20 twenty-minute typing sessions using two different-sized keyboard models. Subjects average over 31 words per minute (WPM) for the first session and increase to an average of 60 WPM by the twentieth. Individual subjects also exceed the upper bound of 60.74 WPM suggested by MacKenzie and Soukoreff's model of two-thumb text entry [5]. We discuss our results in the context of this model.

In a mobile environment, the amount of visual attention a person can devote to a computer is often limited. In addition to typing rapidly and accurately, it is important to be able to enter text with limited visual feedback. Previously we found that users can effectively type in such "blind" conditions with the Twiddler one-handed keyboard. In this paper we examine blind typing on mini-QWERTY keyboards and introduce a taxonomy for blind mobile text Input. We present a study in which eight expert mini-QWERTY typists participated in 5 typing sessions. Each session consists of three twenty minute typing conditions. In the first condition, the control or "normal" condition, the participant had full visual access to both the keyboard and the display. In the second condition, "single blind" we obstructed view of the keyboard. The final "double blind" condition also reduced visual feedback from the display. In contrast to our Twiddler work, we found that in the visually impaired conditions, typing rate and accuracy suffer, never reaching the non-blind rates. Across the blind mini-QWERTY conditions our participants averaged 45.8 wpm at 85.6% accuracy, while blind typing on the Twiddler averaged 47.3 wpm at 93.9% accuracy. We discuss these results in the context of our previous blind typing work and examine the trade-offs between the different keyboards for mobile and wearable computer use.

Even before the advent of the modern windows, icons, menus and pointers (WIMP) interaction paradigm, researchers have focused intently on interfaces which do not chain users to their desks to interact with the same old keyboard, monitor and mouse. The instinct for researchers to explore alternatives to traditional interfaces has been present from the beginning of computing. Indeed, the WIMP paradigm itself was born as an attempt to break away from the text-based, single-threaded computing world that had dominated previously. In defense of the state of the art, many of the limitations on computing interfaces historically resulted from the limitations on the processing capabilities of the computing devices. Moreover, computer scientists and engineers have regularly used additional capacity to improve computing interfaces, showing that the field is at least partially cognizant of the deficiencies in computer interfaces.