基于BERT与DeepSeek大模型的智能舆论监控系统设计

Design of an Intelligent Public Opinion Monitoring System Based on the BERT and DeepSeek Large Model

摘要: 本研究基于BERT模型与DeepSeek大模型，构建了一个智能舆情监测系统。该系统总体架构分为数据采集层、情感分析层、可视化交互层和智能报告层，技术实现上融合了微调BERT模型、Tkinter图形界面以及多源API集成。数据流程涵盖从光明网、Coze等多平台舆情信息采集，到基于公司金融领域微调BERT模型的情感自动标注，再到多维度数据可视化与DeepSeek生成的智能舆情分析报告。系统功能集成舆情动态抓取、情感分类、可视化展示与报告生成四大模块，实现了从数据获取到决策建议的全流程自动化。该系统的建设为舆情监控与风险应对提供了基于深度学习的智能支持，有助于提升企业对突发舆情的响应速度与决策科学性。

Abstract: This study constructed an intelligent public opinion monitoring system based on the BERT and DeepSeek large models. The system’s overall architecture consists of a data collection layer, a sentiment analysis layer, a visualization and interaction layer, and an intelligent reporting layer. Its technical implementation integrates a fine-tuned BERT model, a Tkinter graphical interface, and multi-source API integration. The data pipeline encompasses the collection of public opinion information from multiple platforms, including Guangming.com and Coze, automatic sentiment annotation based on a fine-tuned BERT model for corporate finance, and multi-dimensional data visualization and intelligent public opinion analysis reports generated using DeepSeek. The system integrates four functional modules: dynamic public opinion capture, sentiment classification, visualization, and report generation, automating the entire process from data acquisition to decision-making recommendations. This system provides deep learning-based intelligent support for public opinion monitoring and risk response, helping to improve enterprises’ response speed to sudden public opinion incidents and the scientific nature of their decision-making.