Guo, Shou Xi, et al. "Attention-based character-word hybrid neural networks with semantic and structural information for identifying of urgent posts in MOOC discussion forums." IEEE access 7 (2019): 120522-120532.

This paper introduces a hybrid neural network model designed to identify "urgent" posts in MOOC discussion forums that require immediate attention from instructors. The model combines Convolutional Neural Networks (CNN) and Gated Recurrent Units (GRU) to learn both semantic and structural information from posts, and incorporates Character-level Convolutional Networks (Char-CNN) to address noise such as spelling errors and emoticons. By integrating these features with an attention mechanism, the model achieves high classification performance, with micro F-scores of 91.8%, 91.3%, and 88.4% on the Stanford MOOCPosts dataset, significantly improving upon previous methods. This approach aims to help instructors efficiently manage posts, prioritize responses, and enhance student engagement and course completion rates.