This lightning talk describes our experience using Discord as a platform for online tutoring at the University of Alaska Anchorage (UAA) for the past two and a half semesters. Discord is a free online chat system that is accessible via a web browser, computer program, or mobile app. It competes with apps including Skype, Slack, and TeamSpeak. We initially pitched Discord to our students as a way to deliver online tutoring. However, once a critical mass of students joined the system they began to use it for socializing, finding study partners, and other community-building activities. Utilization has been much higher than other online communication tools we have implemented in the past. With approximately 250 majors there are currently 261 users subscribed to our server with users actively posting messages every day. Students have reported positive experiences with the platform. To further facilitate tutoring in the future we are developing a bot and web tools into the platform to make it easier to share and run snippets of code.

Skilled piano players are able to decipher and play a musical piece they had never seen before (a skill known as sight-reading). For a sample of 23 piano players of various abilities we consider the correlation between machine-extracted gaze path features and the overall human rating. We find that correlation values (between machine-extracted gaze features and overall human ratings) are statistically similar to correlation values between human-extracted task-related ratings (e.g., note accuracy, error rate) and overall human ratings. These high correlation values suggest that an eye tracking-enabled computer could help students assess their sight-reading abilities, and could possibly advise students on how to improve. The approach could be extended to any musical instrument. For keyboard players, a MIDI keyboard with the appropriate software to provide information about note accuracy and timing could complement feedback from an eye tracker to enable more detailed analysis and advice.

The majority of today's authentication systems, including password and fingerprint scanners, are based on one-time, static authentication methods. A continuous, real-time authentication system opens up the possibility for greater security, but such a system must be unobtrusive and secure. In this work we studied whether a commercial eye tracker can be used for unobtrusive, continuous, real-time user authentication via iris recognition. In a user study, all 37 participants could be authenticated with 11% equal error rate (EER). For 14 of the 37 users, iris occlusion was sufficiently small to authenticate with 9% EER. When classified using a k-nearest neighbors algorithm and only the right iris, the same data set allowed 100% accuracy for k = 3. Although these error rates are too high for standalone use, iris recognition via an eye tracker might enable real-time continuous authentication when combined with other more reliable authentication means (e.g., a password). As eye trackers become widely available their capabilities for multiple factor, continuous authentication will become compelling.

Researchers have proposed systems in which users utilize an eye tracker to enter passwords by merely looking at the proper symbols on the computer monitor in the appropriate order. This authentication method is immune to the practice of shoulder surfing: secretly observing the keystrokes of a legitimate user as he or she types a password on a keyboard. In this paper we describe the EyeDent system-in which users authenticate by looking at the symbols on an on-screen keyboard to enter their password. Existing eye-tracking based authentication systems require the user to dwell or press a trigger when looking at each symbol. Instead, in EyeDent, gaze points are automatically clustered to determine the user's selected symbols; this approach has the benefit of allowing users to authenticate at their natural speed, rather than with a fixed dwell time. Additionally, the absence of a visible trigger does not divulge the number of symbols in the password. Results from preliminary investigations indicate that quick (3 seconds for a 4 digit PIN) authentication is possible using this scheme, but more work is needed to account for calibration error, and to dynamically adapt system parameters to the characteristics of individual users.

Graphical passwords are expected to be easier to recall, less likely to be written down and have the potential to provide a richer symbol space than text based passwords. For example, a user might authenticate by clicking a series of points on an image, selecting a series of tiles, or by drawing a series of lines on the screen [Davis et al. 2004]. An example of the tiled approach is the Real User Corporation's PassFaces™ system [Real User, 2005] illustrated in Figure 1. For both text and graphical password entry systems the user needs to carefully enter the password in case a malicious user is observing the session via "shoulder surfing." Although some authors assume that graphical passwords will be entered on a small screen with a reduced observation angle [Jansen 2004], and thus dismiss the likelihood of shoulder surfing, this assumption is not always true.

Many problems are difficult to adequately explore until a prototype exists in order to elicit user feedback. One such problem is a system that automatically categorizes and manages email. Due to a myriad of user interface issues, a prototype is necessary to determine what techniques and technologies are effective in the email domain. This paper describes the implementation of an add-in for Microsoft Outlook 2000 TM that intends to address two problems with email: 1) help manage the inbox by automatically classifying email based on user folders, and 2) to aid in search and retrieval by providing a list of email relevant to the selected item. This add-in represents a first step in an experimental system for the study of other issues related to information management. The system has been set up to allow experimentation with other classification algorithms and the source code is available online in an effort to promote further experimentation.

Many researchers have proposed classification systems that automatically classify email in order to reduce information overload. However, none of these systems are in use today. This paper examines some of the problems with classification technologies and proposes Relevance Categories as a method to avoid some of these problems. In particular, the dynamic nature of email categories, the cognitive overhead, required training categories, and the high costs of classification errors are hurdles for many classification algorithms. Relevance Categories avoid some of these problems through their simplicity; they are merely relevance-ranked lists of email messages that are similar to a set of query messages. by displaying messages as the result of a dynamic query in lieu of fixed categories, we hypothesize that users will be less sensitive to errors using the Relevance Categories scheme than to errors using a fixed categorization scheme. To study the effectiveness of the Relevance Categories concept, we devised a performance metric for relevance ranking and used it to test an inverted index implementation on the Reuter-21578 test collection. The promising test results indicate the need for further work.