DP-FWCA: A Prompt-Enhanced Model for Named Entity Recognition in Educational Domains

The digital transformation in education has spurred a rapid expansion of academic literature, pedagogical resources, and institutional knowledge, intensifying the demand for efficient automated information extraction. However, the inherent complexity and contextual variability of educational texts, compounded by a limited supply of domain-specific annotated data, impose formidable challenges on conventional NER methods. To address these challenges, we propose the Domain-adaptive Prompt Feature-Weighted CNN-Attention-CRF (DP-FWCA), a novel framework specifically designed for educational NER. Our approach integrates domain-adaptive prompting to direct attention toward critical educational semantics, BERT-based contextual embeddings for robust representation learning, multi-scale convolutional neural networks (CNNs) with gated feature fusion to capture fine-grained local features, bidirectional LSTM networks augmented with self-attention to model long-range dependencies, and conditional random fields (CRFs) for structured sequence labeling. Evaluations on the EduNER and MSRA datasets reveal that the DP-FWCA model achieves F1-scores of 87.72% and 95.12%. These results underscore the promise of our integrated approach in overcoming the intrinsic challenges of educational texts, thereby advancing automated knowledge extraction and supporting the development of more intelligent educational systems.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.