Ryan Baker is a Professor at the University of Pennsylvania, and Director of the Penn Center for Learning Analytics. His lab conducts research on engagement and robust learning within online and blended learning, seeking to find actionable indicators that can be used today but which predict future student outcomes. Baker has developed models that can automatically detect student engagement in over a dozen online learning environments, and has led the development of an observational protocol and app for field observation of student engagement that has been used by over 150 researchers in 7 countries. Predictive analytics models he helped develop have been used to benefit over a million students, over a hundred thousand people have taken MOOCs he ran, and he has coordinated longitudinal studies that spanned over a decade. He was the founding president of the International Educational Data Mining Society, is currently serving as Editor of the journal Computer-Based Learning in Context, is Associate Editor of the Journal of Educational Data Mining, was the first technical director of the Pittsburgh Science of Learning Center DataShop, and currently serves as Co-Director of the MOOC Replication Framework (MORF). Baker has co-authored published papers with over 300 colleagues.
Abstract：Contemporary large language models (LLMs) have exploded into much greater use in education in the last year. In this talk, I will discuss ongoing efforts at the Penn Center for Learning Analytics to leverage large language models to support learners and learning, focusing on two projects. In the first project, we use a LLM within a virtual teaching assistant, JeepyTA, who responds to student posts on a discussion forum and gives formative feedback on assignments. In the second project, we use an LLM to improve the depth and quality of feedback given to students learning to program. In this talk, I will discuss the successes and failures of these approaches, and what lessons we can draw from these projects for the broader use of large language models in education.
Jerry Chun-Wei Lin received his Ph.D. from the Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan in 2010. He is currently a full Professor with the Department of Computer Science, Electrical Engineering and Mathematical Sciences, Western Norway University of Applied Sciences, Bergen, Norway. He has published more than 500+ research articles in refereed journals (with 60+ ACM/IEEE transactions journals) and international conferences (IEEE ICDE, IEEE ICDM, PKDD, PAKDD), 16 edited books, as well as 33 patents (held and filed, 3 US patents). His research interests include data mining and analytics, natural language processing (NLP), soft computing, IoTs, bioinformatics, artificial intelligence/machine learning, and privacy preserving and security technologies. He is the Editor-in-Chief of the International Journal of Data Science and Pattern Recognition, the Associate Editor for IEEE TNNLS, IEEE TCYB, IEEE TDSC, INS, JIT, AIHC, IJIMAI, HCIS, IDA, PlosOne, IEEE Access, and the Guest Editor for several IEEE/ACM journals such as IEEE TFS, IEEE TII, IEEE TIST, IEEE JBHI, ACM TMIS, ACM TOIT, ACM TALLIP, and ACM JDIQ. He has recognized as the most cited Chinese Researcher respectively in 2018, 2019, 2020, and 2021 by Scopus/Elsevier. He is the Fellow of IET (FIET), ACM Distinguished Member (Scientist), and IEEE Senior Member.
Abstract：As a large amount of data is collected daily from individuals, businesses and other organizations or applications, various algorithms have been developed to detect interesting and useful patterns in the data that meet a set of requirements specified by a user. The main purpose of data analysis and data mining is to find new, potentially useful patterns that can be used in real-world applications. For example, analyzing customer transactions in a retail store can reveal interesting patterns in customer buying behavior that can then be used for decision-making. In recent years, the demand for utility-based pattern mining and analytics has increased because it can discover more useful and interesting information than simple binary-based pattern mining approaches, which are used in many domains and applications, such as cross-marketing, e-commerce, finance, medicine and biomedicine. In this talk, I will first emphasize the benefits by using the utility-based pattern mining and analysis compared to previous studies (e.g., association rules/frequent itemset mining). Then, I will give a general overview of the state of the art in utility-driven pattern mining and analysis techniques according to three main categories (i.e., data level, constraint level, and application level). Various techniques and modeling on different aspects (levels) of utility-based pattern mining will be presented and reviewed.