This project provides a comprehensive suite for analyzing public opinion from Weibo data, targeting researchers and analysts interested in social media trends. It offers tools for data collection, topic modeling, sentiment analysis, and geospatial visualization, enabling deeper insights into public discourse.

How It Works

The system comprises several modules: a Weibo crawler for comments and user data (approx. 100k/day), LDA topic modeling with data cleaning, segmentation, and optimal topic number selection via coherence and perplexity, and sentiment analysis achieving >97% accuracy. It also includes topic popularity calculation, temporal topic similarity analysis for evolution detection, and map visualizations for sentiment, comment counts, and confirmed cases by province.

Highlighted Details

• Weibo comment and user data crawler (approx. 100k entries per day).
• LDA topic modeling with automated data cleaning, segmentation, and optimal topic number selection.
• Sentiment analysis with reported accuracy exceeding 97%.
• Geospatial visualization of sentiment, comment volume, and case data.

Maintenance & Community

The primary contributor is associated with CSDN and WeChat public accounts "灵海之森" and "西书北影". The author is currently working on large language models, with related open-source repositories available.

Licensing & Compatibility

The repository does not explicitly state a license. Users should verify licensing terms for commercial use or integration into closed-source projects.

Limitations & Caveats

The project is described as a personal research endeavor. Specific details on deployment, dependencies beyond standard Python libraries, and robustness for large-scale production environments are not provided in the README.