Research on the construction of an intelligent evaluation model for enhancing online English writing skills based on multiple factors.

To construct and validate an intelligent evaluation model for online English writing that integrates language proficiency, cognitive strategies, and technology interaction. A three-level indicator system was developed using the Delphi method. Analytic Hierarchy Process (AHP) was employed to derive expert weights, while machine learning extracted behavioral features from student essays and platform logs. A 12-week experiment with 80 non-English major undergraduates was conducted for empirical validation. The model revealed dynamic synergy among the three dimensions. Feedback immediacy and revision frequency significantly predicted writing improvement, while insufficient multimodal resource adaptability emerged as a key limitation. The weight comparison between expert intuition and empirical data showed significant deviations in feedback adoption rate (p < 0.05) and feedback immediacy (p < 0.05), supporting the need for dynamic calibration. The multidimensional model offers a transparent, interpretable framework for process-oriented writing assessment, enabling personalized intervention strategies. The hybrid AHP-ML approach balances human expertise with data-driven insights.