Automatic Coding of Collective Creativity Dialogues in Collaborative Problem Solving Based on Deep Learning Models

Zongxi Li, Haoran Xie, Minhong Wang, Bian Wu, Yiling Hu

Abstract: Creativity and collaboration are considered core competencies of contemporary students in different education levels and disciplines. Existing research mainly focuses on the theoretical framework for computer-supported collaborative learning, and the dialogic content analysis is mainly based on expert annotating. Consequently, there is a vacuum in the direction of AI-based discourse analysis, which prevents researchers from progressing further towards automatic monitoring and assessing collective creativity in problem-solving activities. Hence, this paper aims to fill such a gap by setting a preliminary benchmark for deep learning models in dialogue coding. More concretely, we target identifying metacognition and cognition indicators in a collaborative problem-solving process based on a collective creativity coding framework. Moreover, our work goes beyond the conventional computer-mediated and dyad (one-on-one) settings and focuses on an interactive problem-oriented activity involving multiple participants. We employ deep learning models on the full transcripts collected during the activity to validate the affordance of AI-based coding models in a real teaching and learning scenario. To the best of our knowledge, it is the first attempt to introduce AI techniques into dialogue analysis in collaborative learning.