Reproduce “Predicting States of Understanding in Explanatory Interactions Using Cognitive Load-Related Linguistic Cues” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yu Wang, Olcay Türk, Angela Grimminger, Hendrik Buschmeier
Researchers from Bielefeld and Paderborn Universities developed a multi-class classification model to predict four states of listener understanding during explanatory interactions. The model integrates verbal cues like information value and syntactic complexity, and nonverbal cues such as gaze entropy, achieving a 0.816 average accuracy and 0.812 Macro F1 score when combined with textual features.
Key insights
- Cognitive load-related linguistic cues, specifically information value, syntactic complexity, and gaze entropy, exhibit statistically significant variations across different states of listener understanding.
- Integrating these verbal and nonverbal cues with textual features consistently improves the accuracy and robustness of multi-class classification for predicting understanding states.
- While 'Non-Understanding' is relatively easier to predict, 'Misunderstanding' and 'Partial Understanding' remain more challenging to classify accurately due to inherent complexities and potential confusions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.