Dr. Collin F. Lynch is an internationally-recognized expert on process mining from rich user-system interaction data and data-driven Intelligent Learning Environments (ILEs). He has amassed a strong publication record with 21 journal articles, 5 book chapters 6 edited volumes 100 conference publications. His expertise has been recognized through invited presentations at Hong Kong Polytechnic University the Haifa Center for Law and Technology and the 2019 International Symposium on the Application of AI in Education among others. Dr. Lynch has won competitive funding from the National Science Foundation (NSF) and the Institute for Education Sciences (IES), and has maintained long-term research partnerships with industry and other academic institutions. Dr. Lynch has mentored 15 Ph.D. students 9 of whom have graduated. Four have taken faculty positions, 2 of those are tenure track. He has also supervised 7 M.S. theses and 27 undergraduate researchers. Dr. Lynch has maintained leadership in his research communities through formal roles such as Policy Chair for the IEDMS and co-organization of conferences and workshops, and has developed leadership in the department through his efforts to develop novel degrees in Artificial Intelligence. And he has maintained his commitment to teaching by developing advanced research courses for undergraduate and graduate students, teaching core courses in AI, conducting public outreach through the Osher Lifelong Learning Institute and state agencies. In sum, Dr. Lynch has established an impactful and sustainable research program with a clear path to the future, and he has demonstrated his ongoing commitment to using that research to benefit North Carolina.

Dr Lynch's research program has advanced the state of the art in four primary areas:

1. Rich Data Integration and Process Mining Dr. Lynch has pioneered new techniques for user-system process mining both within and across platforms. This work has been applied to modeling teamwork, help-seeking, self-regulation and problem solving in undergraduate CS courses including assessments of students planning, debugging, struggle, help-seeking, and their ultimate performance in class. More recent work has moved to collecting keystroke-level data to track students' responses to struggle and the intersection between individual practice and group-work. This data is fostering additional research on automated guidance, teamwork, and classroom orchestration.
2. Modeling Reading & Writing Processes and Argumentative Structures Dr. Lynch has developed novel platforms for the collection and analysis of keystroke and mouseclick-level which have been deployed to K-12 classrooms across 4 states. These tools have yielded data reflecting students' writing across disciplines including argumentation and research and their self-regulation in reading and note taking. This work has been paired with the development of models to extract argument and rhetorical structures from text and to model their impact on students. This collection and modeling work has supported the development of instructor-facing dashboards for decision-making and classroom orchestration as well as followup research on explainable AI for writing evaluation.
3. CS Education Dr. Lynch has conducted extensive research on scaffolding, modeling, and developing CS courses as well as interdisciplinary learning. This has included developing ILEs that use embodied cognitive agents, spoken dialogue modeling, and data-driven guidance to support students' coding, collaborative problem-solving, and deliberate practice. It has also included work on modeling student interactions with and responses to undergraduate and graduate coursework within CS and engineering broadly. With competitive support from the NSF and partnerships with existing training programs this has yielded not only technical innovations and pedagogical insights but direct benefits for learners.
4. Policy for Effective AI in Education In order to realize the benefits of data-driven interventions and ILEs it is necessary to ensure that they are effective, auditable, and transparent to students, instructors, and institutions. It is also necessary to ensure that they can be bounded to standards of conversational safety, pedagogy, and reliability. Dr. Lynch has worked to engage with stakeholders in academia and industry to identify shared concerns around AI and to propose policy solutions. This work has yielded 1 journal and 1 conference publication since 2016, and has been supported by Booz Allen Hamilton a longstanding federal contractor with applications in training.