Enhancing Student Dialogue Productivity with Learning Analytics and Fuzzy Rules
Published in 25th International Conference on Artificial Intelligence in Education (AIED24), 2024
Recommended citation: de Araujo, A., Martens, J., & Papadopoulos, P. M. (2024, July). Enhancing Student Dialogue Productivity with Learning Analytics and Fuzzy Rules. In International Conference on Artificial Intelligence in Education (pp. 397-404). Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-64299-9_36 https://link.springer.com/chapter/10.1007/978-3-031-64299-9_36
This study explores the use of the Collaborative Learning Agent for Interactive Reasoning (Clair) in a digital collaborative learning activity where interaction takes place via chat. Clair is designed to adaptively facilitate productive student dialogue using “talk moves” based on the Academically Productive Talk (APT) framework, a popular approach in related conversational agent studies. In this paper, we detail how Clair, powered by learning analytics, machine learning, and a fuzzy rule-based system, can adaptively trigger talk moves in student dialogue. In an experimental study conducted with n = 9 university student dyads, we assess the impact of Clair’s presence on student dialogue productivity. We analyzed the within-subjects differences (with/without Clair) in four key goals of student dialogue productivity: the frequency of (a) students sharing thoughts, (b) orienting and listening, (c) deepening reasoning, and (d) engaging with others’ reasoning. Our findings indicate a notable improvement in deepening reasoning (p = .047), highlighting Clair’s capability to prompt students to engage in more critical thinking and elaborate on their ideas. Yet, the impact on other goals was less pronounced, suggesting the complexity of facilitating all goals of productivity. This paper demonstrates the potential of integrating learning analytics and fuzzy rules into triggering approaches for collaborative conversational agents, offering a novel approach to adaptively trigger talk moves in student dialogue. The results also underline the need for further refinement in the design and application of such systems to comprehensively support productive student dialogues in collaboration settings.
Recommended citation: de Araujo, A., Martens, J., & Papadopoulos, P. M. (2024, July). Enhancing Student Dialogue Productivity with Learning Analytics and Fuzzy Rules. In International Conference on Artificial Intelligence in Education (pp. 397-404). Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-64299-9_36