Publications
Crime prediction plays an impactful role in enhancing public security and sustainable development of urban. With recent advances in data collection and integration technologies, a large amount of urban data with rich crime-related information and fine-grained spatio-temporal logs have been recorded. Such helpful information can boost our understandings of the temporal evolution and spatial factors of urban crimes and can enhance accurate crime prediction. However, the vast majority of existing crime prediction algorithms either do not distinguish different types of crime or treat each crime type separately, which fails to capture the intrinsic correlations among different types of crime. In this paper, we perform crime prediction exploiting the cross-type and spatio-temporal correlations of urban crimes. In particular, we verify the existence of correlations among different types of crime from temporal and spatial perspectives, and propose a coherent framework to mathematically model these correlations for crime prediction. Extensive experiments on real-world datasets validate the effectiveness of our framework.
Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans the biological, physical, and social sciences.
Children who are more curious learn more and perform better in school. Although there has been some work on how to induce curiosity in the moment, much less is known about how to promote curiosity-driven behaviors across contexts. In a preregistered experiment, 103 children ages 5-7-years-old participated in eight virtual science lessons over two weeks. Children were randomly assigned to a condition in which they were encouraged to ask questions, or to a condition in which they were asked to listen carefully. At the end of training, children in the question-asking condition valued new science information significantly more than children in the listening condition. They also learned marginally more than children in the listening condition. Children with less background knowledge benefited more from question-asking practice. Our results suggest that question-asking practice can boost some aspects of curiosity and learning.
Plan recognition is a key component of player modeling. Player plan recognition focuses on modeling how and when players select goals and formulate action sequences to achieve their goals during gameplay. By occasionally asking players to describe their plans, it is possible to devise robust plan recognition models that jointly reason about player goals and action sequences in coordination with player input. In this work, we present a player plan recognition framework that leverages data from player interactions with a planning support tool embedded in an educational game for middle school science education, CRYSTAL ISLAND. Players are prompted to use the planning tool to describe their goals and planned actions in CRYSTAL ISLAND. We use this data to devise data-driven player plan recognition models using multi-label multi-task learning. Specifically, we compare single-task and multi-task learning approaches for both goal prediction and action sequence prediction. Results indicate that multi-task learning yields significant benefits for action sequence prediction. Additionally, we find that incorporating automated detectors of plan completion in plan recognition models improves predictive performance in both tasks.
Some science education researchers have presented either isolated findings on specific points in time during the pandemic or non-empirical insights or suggestions for how teachers, district leaders, policymakers, and others should take up the learnings from the pandemic to move science education forward. However, there are few studies published to date that provide robust and longitudinal empirical data on what science instruction looked like throughout the pandemic and the magnitude of the impacts of the pandemic on science instruction when compared to pre-pandemic science teaching and learning. We conducted a primarily survey-based study on science instruction and enactment of the Next Generation Science Standards (NGSS) in K-8 classrooms throughout the COVID-19 pandemic. This analysis also incorporates a longitudinal dataset from grade 6-8 teachers across California on their NGSS instruction prior to and throughout the first year of the pandemic, providing insight on instruction over multiple years before and throughout distance learning. Our findings highlight the challenges that teachers and students faced during the pandemic, as well as the significant impacts that distance learning appeared to have on science instruction and teachers' ability to provide NGSS-aligned instruction. However, we also found that a year after the initial school closures, teachers' science instruction began to show improvements both in the frequency of science instruction (how often they were able to provide science instruction through distance learning) and the quality of science instruction (how often teachers were able to provide instruction that was aligned with the goals of the NGSS). Implications of this work are far reaching and may impact teachers, students, administrators, policymakers, professional learning providers, and curriculum developers regardless of whether science instruction occurs through distance learning or in-person moving forward.
Research experiences provide educational and intellectual benefits to undergraduate students alongside traditional coursework. However, little prior work has directly examined the relationship between a student's participation in research and their later employment outcomes. Drawing on years of transaction-level administrative records from nine universities, we identify students with paid research experiences on grants from NIH, NSF, and other federal funders alongside a comparison group of students with paid federal work-study employment. Record linkage to later employment information allows us to follow students beyond their university years, measuring estimated salary (n = 15,528) and employment churn (n = 16,628). Using generalized linear models, we estimate differences between research and work-study experiences associated with these two outcomes. Those with federally funded paid research work as undergraduates, our findings demonstrate, have higher salaries and fewer job changes in their early careers. These results support the vision of the university as a site of integrated research and teaching, where students' engagement with research contributes to personal and professional development.
Research indicates that creative cognition depends on both associative and controlled processes, corresponding to the brain's default mode network (DMN) and executive control network (ECN) networks. However, outstanding questions include how the DMN and ECN operate over time during creative task performance, and whether creative cognition involves distinct generative and evaluative stages. To address these questions, we used multivariate pattern analysis (MVPA) to assess how the DMN and ECN contribute to creative cognition over three successive time phases during the production of a single creative idea. Training classifiers to predict trial condition (creative vs non-creative), we used classification accuracy as a measure of the extent of creative activity in each brain network and time phase. Across both networks, classification accuracy was highest in early phases, decreased in mid phases, and increased again in later phases, following a U-shaped curve. Notably, classification accuracy was significantly greater in the ECN than the DMN during early phases, while differences between networks at later time phases were non-significant. We also computed correlations between classification accuracy and human-rated creative performance, to assess how relevant the creative activity in each network was to the creative quality of ideas. In line with expectations, classification accuracy in the DMN was most related to creative quality in early phases, decreasing in later phases, while classification accuracy in the ECN was least related to creative quality in early phases, increasing in later phases. Given the theorized roles of the DMN in generation and the ECN in evaluation, we interpret these results as tentative evidence for the existence of separate generative and evaluative stages in creative cognition that depend on distinct neural substrates.
In the Summer of 2020, as COVID-19 limited in-person research opportunities and created additional barriers for many students, institutions either canceled or remotely hosted their Research Experience for Undergraduates (REU) programs. The present longitudinal qualitative phenomenographic study was designed to explore some of the tools, communities, norms, and division of labor that make up the remote research program and some of the possible limitations, challenges, and outcomes of this remote experience. Within the context of the undergraduate research program, the paper makes a comparison between outcomes of the remote and in-person research experience. Overall, 94 interviews were conducted with paired participants; mentees (N=10) and mentors (N=8) from six different REU programs. By drawing on Cultural-Historical Activity Theory as a framework, our study has explained the remote format has not opened a gap between mentees' research and academic objectives from the REU program and what they got at the end. All mentees reported that this experience was highly beneficial. Comparisons between the outcomes of the remote REU and the published outcomes of in-person UREs revealed many similar benefits of undergraduate research. Our study suggests that remote research programs could be considered as a means to expand access to research experiences for some students even after COVID-19 restrictions are lifted.
Knowledge Tracing (KT) models based on attention mechanisms have demonstrated in literature the capability to predict student performance more accurately than previous models in some datasets. However, they fail to directly infer student knowledge. In this paper, we apply a proposed extension already seen in KT literature in order to infer latent knowledge to these models. We apply the extension to four different attention-based KT models, to investigate whether these models can better infer the knowledge outside the learning system than previous models. We find that attention-based models can generate better knowledge estimate correlations with student's scores than the previous models.
Massive Open Online Courses (MOOCs) are a scalable technology for upskilling but have primarily been successful with highly resourced and well-educated populations. In spite of their low barriers, people experiencing homelessness have not generally benefited from MOOCs. Work-Learn is a new model which utilizes action design research in order to provide individuals experiencing homelessness with the skills and scaffolding that will enable them to benefit from the digital economy. This Work in Progress reports the results of interviews and focus groups focusing on minimum hiring requirements, conducted with eight individuals who are in senior leadership, hiring managers, or training for web development or enterprise computing functions. Preliminary results suggest that after a baseline aptitude has been established, that person-position fit is the critical aspect to consider in instances of homelessness. The mainframe community may represent a unique opportunity due in part to its need for new talent.


