Article Details

Abstract

With the advancement of digital government and the reform of expedited public appeal processing, government service hotlines face dual challenges: accurately processing massive public appeals and deriving deep data insights. To address these, this research designs and implements an intelligent classification and response system for government hotlines based on the pretrained BERT model. The core innovation lies in proposing a hybrid intelligent processing architecture that combines deep semantic understanding with business rule verification and introducing a backend AI analysis agent. Experiments on a real-world government dataset containing 120 000 annotated samples show that the system achieves a macro-averaged F1 score of 91.8% for classification, with 95% of requests complete within an end-to-end response time of under 200 ms. The system significantly improves the processing efficiency and security of government service hotlines, offering a practical technical pathway for the intelligent upgrade of government hotlines and the modernization of governance capacity.

Keywords

government service hotline; BERT model; rule engine; AI agent; intelligent classification

References

1. [1] Chen Shun, Yi Xiuwen, Zhang Junbo, etc Deep Behavioral Semantic Network for 12345 Government Hotline Event Allocation [J]. Computer Science and Exploration, 2025, 19 (4): 1036-1047
2. [2] Zhang Z C. Understanding the Relationship between Normative Records of Appeals and Government Hotline Order Dispatching: A Data Analysis Method [J].Data Technologies and Applications，2024，58（3）：496-516.
3. [3] Li Zichen Research on deep learning based 12345 event deduplication and allocation methods [D]. Chengdu: Southwest Jiaotong University, 2023
4. [4] Devlin J，Chang M-W，Lee K，et al. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding [C]//North American Chapter of the Association for Computational Linguistics: Human Language Technologies，2019：4173-4186.
5. [5] Silva M O，Oliveira G P，Costa L G L，et al. GovBERT-BR: A BERT-based Language Model for Brazilian Portuguese Governmental Data [C]//Intelligent Systems (BRACIS 2024)，2025：19-32.
6. [6] Chen Gang Classification of Government Hotline Work Orders by Integrating RoBERTa and Feature Extraction [J]. Computer and Modernization, 2022 (6): 21-26, 31
7. [7] Xia Peng, Zhang Yijun, Qi Ji Multi agent collaborative code generation technology driven by big language models [J]. Computer Science, 2025, 52 (S2): 771-779
8. [8] Liang Zhumei, Li Bao, Zhao Dongmei Refactoring the Learning Process with AI Agents: Case Analysis and Reflection on the Creation of Teaching Agents [J]. Chinese University Teaching, 2025 (9): 80-86
9. [9] Zeng Weiwei The Risks and Governance of Embedding DeepSeek into the Government Intelligence Ecosystem [J]. Theoretical Exploration, 2025 (5): 37-47
10. [10] Bu Weijue, Zhang Guixian, Zhang Yanmei, etc Overview of Training Methods for Large Language Models: Architecture Evolution, Scaling Laws, and Alignment Techniques [J]. Computer and Network, 2025, 51 (6): 527-534

About the authors

Liu Zhigang (June 1986-), male, Han ethnicity, from Weifang, Shandong, is an engineer and a master's student. His research interests include artificial intelligence, data annotation, and trustworthy data space; Zhao Xiaohui (May 1986-), female, Han ethnicity, from Yuncheng, Shanxi, engineer, master's student, research direction: artificial intelligence, digital economy.