Deep analysis of ideological and political education (IPE) data is vital for enhancing pedagogical precision. Addressing limitations of current methods—particularly in comprehending complex semantics, integrating domain knowledge, and evaluating civic-political (Si-Pol) characteristics—this study proposes an improved deep learning algorithm. Our framework augments semantic understanding by structuring external knowledge, employs a hierarchical attention mechanism to capture key textual information, and incorporates Si-Pol features into the loss function design. Evaluated on the COAE dataset, the model achieved macro-averaged F1 scores of 87.3% and 83.6% on the critical tasks of sentiment orientation recognition and values consistency assessment, respectively. This represents a significant improvement of 3.7% to 8.2% over baseline models (TextCNN, BiLSTM-Attention, base Transformer). The research provides a novel pathway for more accurate and interpretable analysis of IPE effectiveness.
WangXLuC. Qingdao model of educational digitalization: assisting learning, teaching, research, administration, and evaluation. Dig Trans Regi Educ China2025; 1: 451–457.
2.
WangY. Challenges and opportunities of artificial intelligence in ideological and political education in higher education. Inter J High Educ Teach Theo2024; 1: 108.
VaswaniAShazeerNParmarN, et al.Attention is all you need. Adv Neural Inf Process Syst2017; 30: 15.
5.
ChenXWangFLChengG, et al.Understanding learners’ perception of MOOCs based on review data analysis using deep learning and sentiment analysis. Future Internet2022; 14: 218.
6.
KimJJeongCParkS, et al.How do moral emotions shape political participation? a cross-cultural analysis of online petitions using language models. Find Ass Comp Lingu ACL2024; 1: 16274–16289.
7.
RakhlinA. Convolutional neural networks for sentence classification. GitHub2016; 6: 25.
8.
YangZDaiZYangY, et al.Xlnet: generalized autoregressive pretraining for language understanding. Adv Neural Inf Process Syst2019; 32: 1.
9.
YueSWeiJAzizH, et al.A study on the effectiveness of self-assessment learning system of ideological and political education for college students. Learn Motiv2023; 84: 101929.
10.
JangYChoiSJungH, et al.Practical early prediction of students’ performance using machine learning and eXplainable AI. Educ Inf Technol2022; 27: 12855–12889.
11.
BarocasSSelbstAD. Big data’s disparate impact. Calif Law Rev2016; 104: 671.
12.
ShanJ. Opportunities and challenges of ideological and political education in the era of artificial intelligence. Inter J Soc Sci Univ2024; 1: 25.
13.
WeiWSongYYaoB. Enhancing heterogeneous knowledge graph completion with a novel gat-based approach. ACM Trans Knowl Discov Data2024; 18: 1–20.
14.
DongYLiXShiJ, et al.Graph contrastive learning networks with augmentation for legal judgment prediction. Artif Intell Law2024; 5: 1–24.
15.
ShethAGaurMKursuncuU, et al.Shades of knowledge-infused learning for enhancing deep learning. IEEE Internet Comput2020; 23: 54–63.
16.
GaoSYoungMTQiuJX, et al.Hierarchical attention networks for information extraction from cancer pathology reports. J Am Med Inf Assoc2018; 25: 321–330.
17.
ZhouJMaCLongD, et al.Hierarchy-aware global model for hierarchical text classification. Proce 58th Ann Meet Assoc Comp Ling2020; 1: 1106–1117.
18.
PanQZhouJYangD, et al.Mapping knowledge domain analysis in deep learning research of global education. Sustainability2023; 15: 3097.
19.
MutengwaW. Exploring civic education teaching sources, methods and evaluations used by civic education teachers to empower civic education learners with civic skills. Univ Zambia2020; 1: 20.
20.
KatrePD. NLP based text analytics and visualization of political speeches. Int J Recent Technol Eng2019; 8: 8574–8579.
LeopoldEKindermannJ. Text categorization with support vector machines. How to represent texts in input space?Mach Learn2002; 46: 423–444.
23.
ChenY. Convolutional neural network for sentence classification. University of Waterloo2015; 1: 15.
24.
DmoshinskaiaN. Dropout prediction in MOOCs: using sentiment analysis of users' comments to predict engagement. University of Twente2016; 1: 20.
25.
Alva PrincipeRChiariniNVivianiM. Long Document classification in the transformer era: a survey on challenges, advances, and open issues. WIREs Data Min & Knowl2025; 15: e70019.
26.
ÜvegesIRingO. HunEmBERT: a fine-tuned BERT-model for classifying sentiment and emotion in political communication. IEEE Access2023; 11: 60267–60278.
27.
ZhouJCuiGHuS, et al.Graph neural networks: a review of methods and applications. AI Open2020; 1: 57–81.
28.
YangZ-bZhangJ-pZhaoZ-b, et al.Interpreting network knowledge with attention mechanism for bearing fault diagnosis. Appl Soft Comput2020; 97: 106829.
29.
SunMLiPRenJ, et al.Attention mechanism enhanced multi-layer edge perception network for deep semantic medical segmentation. Cognit Comput2023; 15: 348–358.
30.
JinHWangTWanX. Multi-granularity interaction network for extractive and abstractive multi-document summarization. Proce 58th Ann Meet Assoc Comp Ling2020; 1: 6244–6254.
31.
ShanJ. Opportunities and challenges of ideological and political education in the era of artificial intelligence. Inter J Soc Sci Univ2024; 7: 25.
