A Neutrosophic-Rough Set Support for AI-Driven Training Quality Evaluation of Interdisciplinary Communication Talent in the Converged Media Era

Abstract

In the era of converged media, interdisciplinary communication professionals 
must integrate diverse competencies ranging from cross-cultural discourse to multimodal 
content synthesis while adapting to dynamic human–AI collaborative environments. 
Traditional training quality assessment methods fail to capture the complexity, 
uncertainty, and partial contradictions inherent in such contexts. This paper introduces a 
novel Neutrosophic-Rough Set Evaluation Framework (NRSEF) that models training 
performance using a three-dimensional neutrosophic representation: truth (T), 
indeterminacy (I), and falsity (F). Multi-source assessment data from AI analytics, peer 
reviews, and expert evaluations are aggregated using a Choquet-integrated capacity 
measure to account for non-linear competency interactions. The framework’s rough set 
approximation layer provides upper and lower bounds for quality metrics, enabling 
structured treatment of incomplete and conflicting evidence. A case study demonstrates 
the applicability of NRSEF to AI-supported interdisciplinary communication training, 
showing a 23% improvement in predictive accuracy over traditional fuzzy and 
probabilistic methods. The proposed model provides actionable insights for curriculum 
refinement, targeted skill interventions, and adaptive training design in high-uncertainty 
educational environments.

DOI: 10.5281/zenodo.16934636