In this paper we expand on creating dynamically differing animations from single stock animations. This is done using Markov's hierarchy of needs in conjunction with the OCEAN personality model. On start, animations are augmented based on a virtual human agent's parameters from the previous two models. Where Markov's decreases with time, and OCEAN remains static. In our findings this adds more subtle realism to the virtual agent overall and breaks the mirroring effect when agents use the same animations. The subtle changes in the model can be hard to discern unless looking for it. However, we can conclude that this subtle increase in realism has more potential when used in VR settings. Due to the closeness that users will interact with virtual human agents, these subtle changes become more apparent.

As one of the most important and stressful steps towards gaining employment, job interviews are a social barrier that poses a unique challenge for individuals with Autism Spectrum Disorder (ASD). There are existing tools and in-person training available that show promise. However, these solutions are limited in the degree to which they can collect performance data, provide specific feedback, offer options to intelligently customize training, and simulate a variety of interviews in a realistic manner. InterViewR is a mixed reality job interview training simulator that addresses these shortcomings to reduce barriers in entering the workforce. The integration of Virtual Reality and wearable smart technology enables users to practice customizable interview simulations and receive both real-time and retrospective feedback. Findings are reported from a cognitive walkthrough (N=33) that highlights areas that need to be considered when designing such interview training platforms for individuals with autism.

The identification of student struggles has drawn increasing interests from computing education and learning analytics communities in recent years, considering the high failure rate and fast enrollment growth of computer science courses. Prior studies on this topic employed a multitude of data sources and methodologies with varying degrees of success. Nearly all studies attempted to predict low overall course performance to identify struggling students, risking oversimplifying student learning and struggles. Additionally, many studies utilize data sources that are limited to their original contexts or local student demographics, making it difficult to replicate or put the findings into practice. To address these gaps, we studied the feasibility of identifying student struggles at the topic level using features that are agnostic to courses and contexts. Our results show that it is possible to identify student struggles at a more fine-grained level within days. Our findings contribute new insights into automatic identification of student struggles at the topic level on a large scale, which can be used to guide meaningful interventions on student learning.

Job interviews are uniquely challenging for individuals with Autism Spectrum Disorder. While digital interview training tools have shown promising results for improving vocational outcomes for individuals with Autism, existing solutions are largely limited in the degree to which they can simulate a realistic interview environment, collect performance data from the user, and provide actionable feedback for continued improvement. To address these shortcomings and understand how to create an effective training tool, we designed InterViewR, a simulation-based interview training system that combines virtual reality and wearable smart technology in an integrated platform. Our design emphasizes an immersive user experience for training effectiveness and utilizes physiological sensing to provide intelligent affective biofeedback. We report findings from a usability study (N=11) where participants evaluated InterViewR on feasibility, usability, and perceived usefulness.