How can we help users to balance the benefits and risks of information disclosure in a user-friendly manner, so that they can make better privacy decisions? I propose a Privacy Adaptation Procedure that offers tailored decision support. This procedure gives users personalized 'nudges' and 'justifications' based on a context-aware prediction of their privacy preferences.

MOOCs have been very effective at bringing attention to technology and learning. Their focus on remote, asynchronous situations leaves a gap for the co-present, synchronous settings of most university classrooms. I have been looking at the use of technology IN classrooms to better support active student engagement.

This paper describes three exploratory studies on the use of dynamic teams of crowd workers for the rapid com-pletion of open-ended, complex and interdependent projects. We evaluate the design space of expert crowd teams and analyze the unique aspects of these teams, including their elasticity, diversity and time spans.

This paper describes efforts to encourage research experiences among undergraduate students, focusing on endeavors that often lead to large amounts of data but instead resulted in more modest ones. The paper reflects on the lessons that were learned and how efforts to pursue or connect to big data techniques would be helpful.

Many sites designed for software production begin to incorporate social features. In this context, how does the norm of generalized reciprocity work on such platform? Through examining developers' activity records on GitHub, we fount that traditionally defined reciprocity hardly exists. However, a new type of reciprocity ' benefiting others through self-actualization ' starts to emerge.

Children are increasingly at risk for obesity and chronic diseases. One approach is to use ubiquitous technologies to increase health awareness. With recent advances in open source electronics, tangible computing and personal fabrication, children can be empowered to craft their own personal health technologies that can monitor and present data in meaningful ways.

Ancestry.com is an influential intermediary in the online construction of family histories and has considerable influence on the accuracy of family trees being produced by millions of Americans. The website provides access to data on deceased individuals, automated search tools, and tools for the production of family trees. Ancestry's economic success is predicated on providing access to large data repositories filled with data contributed by the company and users. However, little attention is being paid to the accuracy of user-contributed content, leading to the proliferation of inaccurate family trees.

My dissertation aims to examine the value of social question-answering (Q&A) services as platforms for social search in the context of daily lives and investigate how perceptions of credibility and the process of credibility assessment influence people's interactions with other users and with information in social Q&A settings.

We discuss lessons learned from design of a system meant to help NSF personnel understand their funding investments. We term our effort "infrastructure patching "' socio-technical integration of system within a larger infrastructure subsequent to the initial deployment of the original infrastructure. We are happy to demo the system.

While 95% of deaf children in the United States are born to hearing parents, recruiting these parents for research studies is difficult. In this boaster paper, we describe our experiences performing research with this limited target population. We provide suggestions researchers can use to successfully recruit and retain these participants.

This dissertation work seeks to develop a deeper understanding of teens' everyday life interactions with technology by examining the intersections of people, practices, values, and technologies in teens' home contexts. Attention is focused on the spectrums of technology use, including heavy, moderate, low, temporary, and non-use of technologies at home.

We present "Stat!", a new scripting environment for interacting with 'big data': exploratory analytics challenges where the size of the dataset is part of the problem. Stat is an integration between a back-end, a language, environment, and visualization tools designed together to enable progressive queries, collaboration, and sharing.

We investigated collaboration in an instance of evidence-based clinical practice'mental and behavioral health services for children with special needs. Based on our fieldwork, we propose a collaborative information system that helps stakeholders communicate their interpretations of patient data and share relevant knowledge based on their varied areas of expertise.

Analyzing large social media datasets is valuable and difficult. I am building tools and developing methods to scale qualitative research and analysis practices to textual data that challenges traditional methodologies.

Diverse groups of individuals, with their confluence of ideas and perspectives, hold great potential for social creativity. But little is understood about how to harness this creative potential. I highlight ways teams can be constructed and managed to promote good conflict and embrace diverse perspectives while limiting other issues created by diversity.

Despite our understanding of the impact of lifestyle on human health, we lack tools and techniques that capture individuals' behavioral exposures such as diet, sleep and exercise. My current work focuses specifically on capturing eating habits, where I am currently exploring semi-automated food journaling approaches.

Phones, applications, and websites generate vast archives of personal behavioral data. If the data was interconnected and queryable, we could make inferences from it about users, enabling a variety of applications. Imposing a structure on how this data is stored, accessed, and used for inferences will make this possible.

This paper develops the concept of intelligibility in the frame of socio-technical design. Understanding data collected from the users of applications as facilitating dialogue between users and designers, the contingent and ethical nature of design work is discussed. A study of the design of online education is proposed.

Most accessibility work in human-computer interaction focuses on lab-based studies with relatively few participants. A primary and understandable reason is the difficulty of recruiting and accessing specific populations of users. To address this problem, we are exploring alternate methods to increase the scale at which we conduct accessibility research, turning to online study participation and analysis of user-generated content to both understand existing technology use and evaluate new technologies. As one example, we recently collected and analyzed 187 non-commercial YouTube videos depicting users with motor impairments interacting with mainstream touchscreen devices. This analysis yielded a richer characterization of use than would have been possible in a small-scale lab study, and demonstrated the utility of mining user-generated content to study user interface design.

Understanding self biometrically provides users with a view of their lifestyle that they have never seen before. EEG data can be used to quantify the amount of time a person spends 'cognitively engaged' in activities throughout the day. This process creates gigabytes of 'excessive' data. What essential bits really matter to the user, and how do we find out?

Although pedestrian maps/routing algorithms continue to improve, there is currently no way for a user to determine the accessibility of city areas prior to travel. To address this, we present early explorations of applying computer vision techniques to automatically detect curb ramps in Google Street View imagery.

We're currently experiencing an "online learning revolution" -- but what about the 65% of the world that can't take advantage of it? KA Lite is a lightweight web app for serving Khan Academy content (videos and exercises) without needing internet connectivity, from a local server such as a Raspberry Pi.

Network analysis requires understanding communities, but standard approaches for showing them do not sufficiently expose details and relationships. We propose two Group-in-a-Box layouts that display each community within its own appropriately sized box. An evaluation with Twitter networks demonstrates the utility of the layouts for showing communities and their relationships.

Self-monitoring is one of the core drivers of behavior change. Tracking sleep and lifestyle factors could bring people's awareness to the previously unrecognized problems. We explore ways to help people capture sleep and sleep-related factors with a low burden and to increase the awareness of the factors affecting their sleep.

Although there are dozens of community detection algorithms, we lack real community structures necessary to create reliable benchmarks for evaluation of such clustering algorithms. We developed a Facebook application to explore three common community detection algorithms. With this tool, we learn from people and real networks to evaluate existing approaches.

Reaching consensus is a desired but elusive goal in many distributed discussions. A fundamental problem is that the communication interfaces used for the discussions (e.g. Web forums) lack mechanisms for building consensus. In this paper we introduce Procid, a novel interactive system supporting consensus building in distributed design discussions.

I work on crowdsourcing subjective tasks: tasks that computers can't do because they involve language abilities, cultural understanding and understanding people. Currently, I am crowdsourcing taxonomy creation. Taxonomies are a subjective organization of data based on what users find interesting. My current system, Frenzy, uses friendsourcing to collaboratively create taxonomies.

Many documents contain charts to visually reinforce the text. However, finding the marks in the chart that refer to text can be challenging. We present an initial foray into extracting refer- ences in text to charts using crowdsourcing, and describe methods for combining crowdsourced references and measures for assessing reference quality.

We live in a world of 'planned obsolescence' where a product's lifetime is predetermined on purpose. Because of that, people nowadays take a non-repairable electronics for granted. This paper explores a crowdsourcing platform to co-create and share repair manuals for electronics to help promote DIY electronics repairing practices.

My research focuses on promoting a better understanding and more effective use of machine learning techniques in large-scale data analysis. I develop visual analytics tools for evaluating, designing, and deploying statistical models, to ensure that the models are responsive to analysis needs and accessible to a general audience.

We explore an increasingly feasible technological concept called House Memory. This concept invites you to think about living in a charming old house that 'remembers' how and why it was repaired, altered, or improved by previous owners, and think about how the house's memory might help you better understand how it works and how you should maintain it.