Article: Campus network public opinion monitoring method based on emotional feature extraction and classification Journal: International Journal of Continuing Engineering Education and Life-Long Learning (IJCEELL) 2026 Vol.36 No.11 pp.126 - 150 Abstract: This study proposes a multimodal approach for timely campus network public opinion monitoring amidst increasing data. The method uses convolutional neural networks, bidirectional encoders, and Mel-frequency cepstral coefficients to extract features from images, text, and speech, which are then fused using a bidirectional gated recurrent unit (BiGRU) for sentiment classification. Comparative experiments demonstrated the superiority of the BiGRU and self-attention mechanism algorithm, achieving an F1 score of 0.85 and accuracy of 0.87. This method consistently achieved the highest accuracy, averaging close to 0.90, with minimal variation as sample size increased. It also maintained the shortest, most stable response time. The proposed monitoring method demonstrates significant performance and high accuracy, offering valuable support for campus management. Inderscience Publishers - linking academia, business and industry through research

Title: Campus network public opinion monitoring method based on emotional feature extraction and classification

Authors: Xue Li; Shanshan Li; Yunge Gao; Kang Liu

Addresses: Faculty of Humanities and Social Sciences, SanYa Aavation and Tourism College, Sanya, 572000, China ' Faculty of Humanities and Social Sciences, SanYa Aavation and Tourism College, Sanya, 572000, China ' Faculty of Humanities and Social Sciences, SanYa Aavation and Tourism College, Sanya, 572000, China ' Faculty of Humanities and Social Sciences, SanYa Aavation and Tourism College, Sanya, 572000, China

Abstract: This study proposes a multimodal approach for timely campus network public opinion monitoring amidst increasing data. The method uses convolutional neural networks, bidirectional encoders, and Mel-frequency cepstral coefficients to extract features from images, text, and speech, which are then fused using a bidirectional gated recurrent unit (BiGRU) for sentiment classification. Comparative experiments demonstrated the superiority of the BiGRU and self-attention mechanism algorithm, achieving an F1 score of 0.85 and accuracy of 0.87. This method consistently achieved the highest accuracy, averaging close to 0.90, with minimal variation as sample size increased. It also maintained the shortest, most stable response time. The proposed monitoring method demonstrates significant performance and high accuracy, offering valuable support for campus management.

Keywords: bidirectional encoder representations from transformers; multi-modal features; Mel-frequency cepstral coefficients; MFCC; bidirectional gated recurrent unit; GRU; campus network public opinion monitoring.

DOI: 10.1504/IJCEELL.2026.154667

International Journal of Continuing Engineering Education and Life-Long Learning, 2026 Vol.36 No.11, pp.126 - 150

Received: 16 Oct 2025
Accepted: 10 Feb 2026
Published online: 09 Jul 2026 *

Title: Campus network public opinion monitoring method based on emotional feature extraction and classification

Keep up-to-date