CAREER: Towards Intelligent Learning Environments that Support the Practice of Programming

点击翻译按钮获取中文摘要

Technology plays an important role in modern society, shaping how people socialize, learn, and work. As a result, programming skills are in high demand across all sectors of the economy. While enrollments in university computer science (CS) courses are growing rapidly, many students struggle to learn programming and retention in the major is low. Learning to program requires mastering practices like problem-solving, systematic debugging, and adaptive planning. However, rising enrollments in CS classes make it difficult for instructors to monitor student practices and provide feedback. Given that a student’s programming process cannot be determined from the final solution, instructors rarely have visibility into students’ programming practices. This project leverages a unique opportunity to observe and support the programming process automatically and aims to advance scientific understanding of the programming process by building intelligent programming environments that assess and adapt to students’ motivations and practices. As a CAREER project, it employs integrated education and outreach efforts, to develop the theoretical and technical foundations needed to develop and disseminate intelligent learning environments that effectively support the practice of programming.This CAREER project explores and addresses core challenges in promoting effective programming practices in introductory CS courses. The PI’s prior work reveals that students often use programming practices as signals of whether they are performing well. For example, many students believe that planning and looking up syntax are signs of low ability, even though experts consider these practices a natural part of programming. Furthermore, these self-assessments have been shown to impact self-efficacy, or a student’s belief in their ability to succeed. These findings introduce two core challenges. First, students’ inaccurate expectations about programming may lead them to avoid effective expert practices. Second, students may develop low self-efficacy when they engage in these practices, a factor that can impact course performance and the decision to major in CS. This project proposes a general approach and set of techniques to address these challenges by (1) developing behavioral models that automatically detect programming practices to enable a large-scale study of student motivations and practices, and (2) designing and evaluating interventions that provide personalized guidance to help students develop motivation and effective practices. The proposed education plan will expand the impact of the research through (1) online workshops for CS instructors that focus on student motivation and practices, (2) new courses for graduate students who plan to become CS instructors, and (3) research opportunities for students who are underrepresented in CS.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.