Hybrid Multimodal Fusion Assessment of Early Childhood Teachers' Observational Competence in Mobile and Blended Learning
Year: 2026
Type: article
Abstract: Early Childhood Teachers (ECTs)' observational competence underpins child-centered preschool education. However, traditional assessment approaches face key bottlenecks: high subjectivity, poor scalability, and failure to capture implicit cognitive processes in mobile and blended teacher professional-development contexts. Existing multimodal research lacks targeted, adaptive evaluation systems for this educational setting. This study proposes a hybrid feature-level + decision-level multimodal fusion framework to build a dual-output assessment model. Leveraging six-month longitudinal data from 78 teachers, the model integrates video, audio, text, and behavioral log data. Empirical results show 88.6 % assessment accuracy and a 0.91 correlation with expert gold-standard ratings, with better observed performance than single-modal and single-layer fusion alternatives. This work provides an extensible, objective tool for data-driven precision teacher development toward prospective mobile and blended-learning-oriented professional training.
DOI: 10.4018/ijmbl.427272
Author: Yuan Ma
Institution: Changsha Normal University
Language: Ndonga
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7225376449