Sub-theme: Designing and evaluating adaptive and personalized user interfaces

Adaptive Click and Cross: Adapting to Both Abilities and Task to Improve Performance of Users With Impaired Dexterity

Adaptive Click-and-Cross, an interaction technique for computer users with impaired dexterity. This technique combines three "adaptive" approaches that have appeared separately in previous literature: adapting the user's abilities to the interface (i.e., by modifying the way that the cursor works), adapting the user interface to the user's abilities (i.e., by modifying the user interface through enlarging items), and adapting the user interface to the user's task (i.e., by moving frequently or recently used items to a convenient location). Adaptive Click-and-Cross combines these three adaptations to minimize each approach's shortcomings, selectively enlarging items predicted to be useful to the user while employing a modified cursor to enable access to smaller items.

Exploring The Design Space Of Adaptive User Interfaces

For decades, researchers have presented different adaptive user interfaces and discussed the pros and cons of adaptation on task performance and satisfaction. Little research, however, has been directed at isolating and understanding those aspects of adaptive interfaces which make some of them successful and others not. We have conducted several laboratory studies to systematically isolate some of the design and contextual factors that affect the impact of adaptation on users' performance and satisfaction. The results of these studies combined with the recent work of others, provide an initial characterization of the design space of adaptive graphical user interfaces.

SPRWeb: Preserving Subjective Responses to Website Colour Schemes through Automatic Recolouring

Colors are an important part of user experiences on the Web. Color schemes influence the aesthetics, first impressions and long-term engagement with websites. However, five percent of people perceive a subset of all colors because they have color vision deficiency (CVD), resulting in an unequal and less-rich user experience on the Web. Traditionally, people with CVD have been supported by recoloring tools that improve color differentiability, but do not consider the subjective properties of color schemes while recoloring. To address this, we developed SPRWeb, a tool that recolors websites to preserve subjective responses and improve color differentiability, thus enabling users with CVD to have similar online experiences. SPRWeb is the first tool to automatically preserve the subjective and perceptual properties of website color schemes thereby equalizing the color-based web experience for people with CVD.

SUPPLE: Automatically Generating User Interfaces

SUPPLE is an application- and device-independent system that automatically generates user interfaces for a wide variety of display devices. SUPPLE uses decision-theoretic optimization to render an interface from an abstract functional specification and an interchangeable device model. SUPPLE can use information from the user model to automatically adapt user interfaces to different tasks and work styles while also prividing extensive customization mechanisms that allow for modifications to the appearance, organization and navigational structure of the user interface.

ARNAULD: Preference Elicitation For Interface Optimization

Recent years have revealed a trend towards increasing use of optimization as a method for automatically designing aspects of an interface's interaction with the user. In most cases, this optimization may be thought of as decision-theoretic -- the objective is to minimize the expected cost of a user's interactions or (equivalently) to maximize the user's expected utility. While decision-theoretic optimization provides a powerful, flexible, and principled approach for these systems, the quality of the resulting solution is completely dependent on the accuracy of the underlying utility or cost function. Unfortunately, determining the correct utility function is a complex, time-consuming, and error-prone task. While domain-specific learning techniques have been used occasionally, most practitioners parameterize the utility function and then engage in a laborious and unreliable process of hand-tuning.

Sub-theme: Characterizing individual differences that interfaces ought to adapt to

Exploring Individual Differences in How People Use Adaptive User Interfaces

Past work, including ours, has shown that well-designed adaptive user interfaces can substantially improve people's performance and that people prefer those interfaces to the standard one-size-fits-all designs. But do all people benefit from adaptive user interfaces equally, or are some systematic differences causing some people to reap greater benefit than others. Our first study, which utilized the results from the Multitasking Test on Lab in the Wild has shown that people with high need for cognition utilize the adaptive feature fo adaptive user interfaces much more than those with low need for cognition. Also, introverts utilize the adaptive interface more than extroverts.

Predicting Users' First Impressions of Website Aesthetics

Users make lasting judgments about a website's appeal within a split second of seeing it for the first time. This first impression is influential enough to later affect their opinion of a site's usability and trustworthiness. In this project, we aim to automatically adapt website aesthetics to users' various preferences in order to improve this first impression. As a first step, we are working on predicting what people find appealing, and how this is influenced by their demographic backgrounds.

Cultural Differences in Time Perception and Group Decision-Making

When discussing the effect of technology on culture, people often assume that the world is slowly homogenizing into a culture of Internet users, who share similar values and behavioral norms. Our analysis of the online scheduling behavior on Doodle argues against this hypothesis. In fact, event scheduling is not simply a matter of finding a mutually agreeable time, but a process that is shaped by social norms and values. And this can highly vary between countries. To investigate the influence of national culture on people's scheduling behavior we analyzed more than 1.5 million Doodle date/time polls from 211 countries. Our findings include that people around the world steer their availabilities towards those options that have good chances to reach consensus. But people from more group-oriented collectivist countries (e.g., India, China) seem to make a larger effort to reach mutual agreement than individualists (e.g., the US). We believe that increasing the awareness of such differences can help improve intercultural scheduling and support the acceptance of cultural differences as an interesting contribution to our lives.

Goal crossing vs area pointing for people with motor impairments

Prior work has highlighted the challenges faced by people with motor impairments when trying to acquire on-screen targets using a mouse or trackball. Two reasons for this are the difficulty of positioning the mouse cursor within a confined area, and the challenge of accurately executing a click. We hypothesize that both of these difficulties with area pointing may be alleviated in a different target acquisition paradigm called "goal crossing." In goal crossing, users do not acquire a confined area, but instead pass over a target line. Although goal crossing has been studied for able-bodied users, its suitability for people with motor impairments is unknown. In our study, participants with motor impairments were faster with and preferred goal-crossing to area pointing. This work provides the empirical foundation from which to pursue the design of crossing-based interfaces as accessible alternatives to pointing-based interfaces.

Lab in the Wild

Lab in the Wild is a platform for conducting large scale behavioral experiments with unpaid online volunteers. LabintheWild helps make empirical research in Human-Computer Interaction more reliable (by making it possible to recruit many more participants than would be possible in conventional laboratory studies) and more generalizable (by enabling access to very diverse groups of participants).

LabintheWild experiments typically attract thousands or tens of thousands of participants (with two studies reaching more than 250,000 people). LabintheWild's volunteer participants have also been shown to provide more reliable data and exert themselves more than participants recruited via paid platforms (like Amazon Mechanical Turk). A key characteristic of LabintheWild is its incentive structure: Instead of money, participants are rewarded with information about their performance and an ability to compare themselves to others. This design choice engages curiosity and enables social comparison---both of which motivate participants.

LabintheWild is co-directed by Profs. Katharina Reinecke and Krzysztof Gajos.

Understanding Crowdsourcing as a Tool for Research

Many excellent crowdsourcing and citizen science tools exist to support research, but only a tiny fraction of researchers make use of them. Why? Our interviews with 18 researchers across disciplines revealed a number of differences between the assumptions underlying existing crowd- and citizen-powered platforms and the prevalent research practices and norms.

Curiosity

We built on the information gap theory of curiosity to develop several interventions to motivate crowdworkers to persist longer on a task. Our experiment results show that curiosity interventions improve worker retention without degrading performance, and the magnitude of the effects are influenced by both the personal characteristics of the worker and the nature of the task.

Crowdsourcing Performance Evaluations of User Interfaces

Can computer users be trusted to paricipate in user interface studies from the comfort of their home? Can user interface researchers give up control over their subjects' environment? In this project we study whether we can use Amazon Mechanical Turk to conduct user interface studies reliably. To do so, we replicated three previously known performance experiments, the "Bubble Cursor," the "Split Menus," and the "Split Interface," both in our lab and on Mechanical Turk. We compared the lab with the online population in terms of performance metrics such as speed, accuracy, and consistency. The results show that the Mechanical Turk participants perform just as well as the lab participants.

DERBI: Communicating Individual Biomonitoring and Personal Exposure Results to Study Participants

Epidemiologic studies and public health biomonitoring rely on chemical exposure measurements in blood, urine, and other tissues, and in personal environments, such as homes. For many chemicals, the health implications of individual results are uncertain, and the sources and strategies to reduce exposure may not be known. Yet, a growing number of researchers consider it their ethical obligation to report the results back to their participants. In a project led by the Silent Spring Institute, we are building scalable online tools to help researchers communicate personalized results to study participants in a manner that appropriately conveys what is and what is not known about the sources and effects of different environmental chemicals.

Accurate Measurements of Pointing Performance from In Situ Observations

We present a method for obtaining lab-quality measurements of pointing performance from unobtrusive observations of natural in situ interactions. Specifically, we have developed a set of user-independent classifiers for discriminating between deliberate, targeted mouse pointer movements and those movements that were affected by any extraneous factors. Our results show that, on four distinct metrics, the data collected in-situ and filtered with our classifiers closely matches the results obtained from the formal experiment.

Curio: a platform for crowdsourcing research tasks in sciences and humanities

Curio is intended to be a platform for crowdsourcing research tasks in sciences and humanities. The platform is designed to allow researchers to create and launch a new crowdsourcing project within minutes, monitor and control aspects of the crowdsourcing process with minimal effort. With Curio, we are exploring a brand new model of citizen science that significantly lowers the barrier of entry for scientists, developing new interfaces and algorithms for supporting mixed-expertise crowdsourcing, and investigating a variety of human computation questions related to task decomposition, incentive design and quality control.

Supporting Effective Collective Ideation at Scale

Various online platforms for different domains--ranging from social development to product design--have emerged as a space where people can share their ideas and get inspired by ideas from other people all over the world. The promise of these platforms is that the mix of perspectives and expertise among the participants should allow creative solutions to emerge in ways unimaginable in the lone-innovator or small-group settings. In practice, however, existing online innovation platforms accumulate large numbers of mundane and repetitive ideas rarely leading to valuable breakthroughs.

We have developed IdeaHound, an online platform that helps large groups of people generate diverse ideas together. IdeaHound is enabled by a crowd- and machine learning-based technique to generate a computational representation of the solution space, called an idea map, that encodes semantic relationships between ideas. The results of a subsequent study show that by presenting an automatically sampled set of creative and diverse example ideas from the idea map, IdeaHound can improve the diversity and creativity of ideas generated by a participants compared to presenting a set of randomly selected examples. A subsequent study shed light on the best timing for delivery of inspirational examples.

Curiosity

We built on the information gap theory of curiosity to develop several interventions to motivate crowdworkers to persist longer on a task. Our experiment results show that curiosity interventions improve worker retention without degrading performance, and the magnitude of the effects are influenced by both the personal characteristics of the worker and the nature of the task.

Organic Peer Assessment

We are developing tools and techniques for organic peer assessment, an approach where assessment occurs as a side effect of students performing activities, which they find intrinsically motivating. Our preliminary results, obtained in the context of a flipped classroom, show that the quality of the summative assessment produced by the peers matched that of experts, and we encountered strong evidence that our peer assessment implementation had positive effects on achievement.

Learnersourcing: Leveraging Crowds of Learners to Improve the Experience of Learning from Videos

Rich knowledge about the content of educational videos can be used to enable more effective and more enjoyable learning experiences. We are developing tools that leverage crowds of learners to collect rich meta data about educational videos as a byproduct of the learners' natural interactions with the videos. We are also developing tools and techniques that use these meta data to improve the learning experience for others.

Crowdsourcing Performance Evaluations of User Interfaces

Can computer users be trusted to paricipate in user interface studies from the comfort of their home? Can user interface researchers give up control over their subjects' environment? In this project we study whether we can use Amazon Mechanical Turk to conduct user interface studies reliably. To do so, we replicated three previously known performance experiments, the "Bubble Cursor," the "Split Menus," and the "Split Interface," both in our lab and on Mechanical Turk. We compared the lab with the online population in terms of performance metrics such as speed, accuracy, and consistency. The results show that the Mechanical Turk participants perform just as well as the lab participants.

Curio: a platform for crowdsourcing research tasks in sciences and humanities

Curio is intended to be a platform for crowdsourcing research tasks in sciences and humanities. The platform is designed to allow researchers to create and launch a new crowdsourcing project within minutes, monitor and control aspects of the crowdsourcing process with minimal effort. With Curio, we are exploring a brand new model of citizen science that significantly lowers the barrier of entry for scientists, developing new interfaces and algorithms for supporting mixed-expertise crowdsourcing, and investigating a variety of human computation questions related to task decomposition, incentive design and quality control.

Mobi: Human Computation Tasks with Global Constraints

An important class of tasks that are underexplored in current human computation systems are complex tasks with global constraints. One example of such a task is itinerary planning, where solutions consist of a sequence of activities that meet requirements specified by the requester. In this paper, we focus on the crowdsourcing of such plans as a case study of constraint-based human computation tasks and introduce a collaborative planning system called Mobi that illustrates a novel crowdware paradigm. Mobi presents a single interface that enables crowd participants to view the current solution context and make appropriate contributions based on current needs. We conduct experiments that explain how Mobi enables a crowd to effectively and collaboratively resolve global constraints, and discuss how the design princi- ples behind Mobi can more generally facilitate a crowd to tackle problems involving global constraints.

PlateMate: Crowdsourcing Nutrition Analysis from Food Photographs

PlateMate allows users to take photos of their meals and receive estimates of food intake and composition. Accurate awareness of this information is considered a prerequisite to successful change of eating habits, but current methods for food logging via self-reporting, expert observation, or algorithmic analysis are time-consuming, expensive, or inaccurate. PlateMate crowdsources nutritional analysis from photographs using Amazon Mechanical Turk, automatically coordinating untrained workers to estimate a meal's calories, fat, carbohydrates, and protein. To make PlateMate possible, we developed the Management framework for crowdsourcing complex tasks, which supports PlateMate's decomposition of the nutrition analysis workflow. Two evaluations show that the PlateMate system is nearly as accurate as a trained dietitian and easier to use for most users than traditional self-reporting, while remaining robust for general use across a wide variety of meal types.

Automatic Task Design on Amazon Mechanical Turk

A central challenge in human computation is in understanding how to design task environments that effectively attract participants and coordinate the problem solving process. We consider a common problem that requesters face on Amazon Mechanical Turk: how should a task be designed so as to induce good output from workers? In posting a task, a requester decides how to break down the task into unit tasks, how much to pay for each unit task, and how many workers to assign to a unit task. These design decisions affect the rate at which workers complete unit tasks, as well as the quality of the work that results. Using image labeling as an example task, we consider the problem of designing the task to maximize the number of quality tags received within given time and budget constraints. We consider two different measures of work quality, and construct models for predicting the rate and quality of work based on observations of output to various designs. Preliminary results show that simple models can accurately predict the quality of output per unit task, but are less accurate in predicting the rate at which unit tasks complete. At a fixed rate of pay, our models generate different designs depending on the quality metric, and optimized designs obtain significantly more quality tags than baseline comparisons.

TELLab: Experimentation @ Scale to Support Experiential Learning in Social Sciences and Design

Well-conducted lecture demonstrations of natural phenomena improve students' engagement, learning and retention of knowledge. Similarly, laboratory modules that allow for genuine exploration and discovery of relevant concepts can improve learning outcomes. These pedagogical techniques are used frequently in natural sciences and engineering to teach students about phenomena in the physical world. But how might we conduct a lecture demonstration to demonstrate impact of extraneous cognitive load on performance? How might we design a lab, in which students explore how adding decorations to visualizations impacts the comprehension and memorability of visualizations? We are developing tools, content and procedures to bring experiential learning techniques to social science and design-related courses that teach concepts related to human perception, cognition and behavior. Specifically, we are working to develop software technologies to enable rapid, large-scale and ethical online human-subjects experimentation in undergraduate design-related courses. See the project web site for more.

Organic Peer Assessment

We are developing tools and techniques for organic peer assessment, an approach where assessment occurs as a side effect of students performing activities, which they find intrinsically motivating. Our preliminary results, obtained in the context of a flipped classroom, show that the quality of the summative assessment produced by the peers matched that of experts, and we encountered strong evidence that our peer assessment implementation had positive effects on achievement.

Learnersourcing: Leveraging Crowds of Learners to Improve the Experience of Learning from Videos

Rich knowledge about the content of educational videos can be used to enable more effective and more enjoyable learning experiences. We are developing tools that leverage crowds of learners to collect rich meta data about educational videos as a byproduct of the learners' natural interactions with the videos. We are also developing tools and techniques that use these meta data to improve the learning experience for others.

Ingenium: Improving Engagement and Accuracy with the Visualization of Latin for Language Learning

Learners commonly make errors in reading Latin, because they do not fully understand the impact of Latin's grammatical structure--its morphology and syntax--on a sentence's meaning. Synthesizing instructional methods used for Latin and artificial programming languages, Ingenium visualizes the logical structure of grammar by making each word into a puzzle block, whose shape and color reflect the word's morphological forms and roles. See the video to see how it works.

PETALS Project -- A Visual Decision Support Tool For Landmine Detection

We have built an interactive visualization system, callded PETALS, that helps novice deminers learn how to correctly identify dangerous cluster configurations of landmines. We conducted a controlled study with two experienced instructors from the Humanitarian Demining Training Center (HDTC) in Fort Leonard Wood, Missouri and 58 participants, who were put through the basic land mine detection course. Half of the participants had access to PETALS during training and half did not. During the final exam, which all participants completed without PETALS, participants who used PETALS during training were 72% less likely to make a mistake on the cluster tasks. These results are not yet published, but the available papers capture the initial development and evaluation of the PETALS system.

Model of Technology Acceptance by Older Adults

Mobile technologies have the potential to improve the quality of life of older adults by supporting self-management of cronic care, providing ways to maintain social bonds with family and community, by facilitating access to community resources, and through a myriad of other ways. Older adults, however, are less likely to adopt mobile technologies than younger adults. Why? So far, we have found one key factor not accounted for by other technology acceptance models. See the paper for more:

SPRWeb: Preserving Subjective Responses to Website Colour Schemes through Automatic Recolouring

Colors are an important part of user experiences on the Web. Color schemes influence the aesthetics, first impressions and long-term engagement with websites. However, five percent of people perceive a subset of all colors because they have color vision deficiency (CVD), resulting in an unequal and less-rich user experience on the Web. Traditionally, people with CVD have been supported by recoloring tools that improve color differentiability, but do not consider the subjective properties of color schemes while recoloring. To address this, we developed SPRWeb, a tool that recolors websites to preserve subjective responses and improve color differentiability, thus enabling users with CVD to have similar online experiences. SPRWeb is the first tool to automatically preserve the subjective and perceptual properties of website color schemes thereby equalizing the color-based web experience for people with CVD.

Adaptive Click and Cross: Adapting to Both Abilities and Task to Improve Performance of Users With Impaired Dexterity

Adaptive Click-and-Cross, an interaction technique for computer users with impaired dexterity. This technique combines three "adaptive" approaches that have appeared separately in previous literature: adapting the user's abilities to the interface (i.e., by modifying the way that the cursor works), adapting the user interface to the user's abilities (i.e., by modifying the user interface through enlarging items), and adapting the user interface to the user's task (i.e., by moving frequently or recently used items to a convenient location). Adaptive Click-and-Cross combines these three adaptations to minimize each approach's shortcomings, selectively enlarging items predicted to be useful to the user while employing a modified cursor to enable access to smaller items.

SUPPLE: Automatically Generating User Interfaces

SUPPLE is an application- and device-independent system that automatically generates user interfaces for a wide variety of display devices. SUPPLE uses decision-theoretic optimization to render an interface from an abstract functional specification and an interchangeable device model. SUPPLE can use information from the user model to automatically adapt user interfaces to different tasks and work styles while also prividing extensive customization mechanisms that allow for modifications to the appearance, organization and navigational structure of the user interface.

Goal crossing vs area pointing for people with motor impairments

Prior work has highlighted the challenges faced by people with motor impairments when trying to acquire on-screen targets using a mouse or trackball. Two reasons for this are the difficulty of positioning the mouse cursor within a confined area, and the challenge of accurately executing a click. We hypothesize that both of these difficulties with area pointing may be alleviated in a different target acquisition paradigm called "goal crossing." In goal crossing, users do not acquire a confined area, but instead pass over a target line. Although goal crossing has been studied for able-bodied users, its suitability for people with motor impairments is unknown. In our study, participants with motor impairments were faster with and preferred goal-crossing to area pointing. This work provides the empirical foundation from which to pursue the design of crossing-based interfaces as accessible alternatives to pointing-based interfaces.

Opportunity Knocks: a System to Provide Cognitive Assistance with Transportation Services

Opportunity Knocks (OK) is an automated transportation routing system, whose goal is to improve the efficiency, safety and independence of individuals with mild cognitive disabilities. OK is implemented on a combination of a Bluetooth sensor beacon that broadcasts GPS data, a GPRS-enabled cell-phone, and remote activity inference software. The system uses a novel inference engine that does not require users to explicitly provide information about the start or ending points of their journeys; instead this information is learned from users' past behavior.

Accurate Measurements of Pointing Performance from In Situ Observations

We present a method for obtaining lab-quality measurements of pointing performance from unobtrusive observations of natural in situ interactions. Specifically, we have developed a set of user-independent classifiers for discriminating between deliberate, targeted mouse pointer movements and those movements that were affected by any extraneous factors. Our results show that, on four distinct metrics, the data collected in-situ and filtered with our classifiers closely matches the results obtained from the formal experiment.