TikTok Text Sentiment and Analytics Dashboard
Python · PostgreSQL · Google Gemini · Streamlit · Plotly · GitHub Actions
Overview
This project analyzes TikTok-related text that is already stored in a PostgreSQL table. A scheduled Python job selects unanalyzed rows, asks the Gemini API to classify sentiment and produce a short Polish summary, then saves the result. A Streamlit dashboard presents the analyzed records and engagement metrics.
Workflow
- Read pending text and post metadata from PostgreSQL.
- Classify sentiment as positive, neutral, or negative and generate a one-sentence summary in Polish with Google Gemini.
- Write the analysis back to PostgreSQL, with a delay and handling for API rate limits.
- Run the agent on a GitHub Actions schedule or manually.
- Explore sentiment distribution, popular posts, and filtered records in the Streamlit dashboard.
Technologies
Python, PostgreSQL, psycopg2, Google Gemini API, python-dotenv, Streamlit, Pandas, Plotly Express, GitHub Actions.
Scope
The repository processes text already present in the database; it does not collect or scrape content from TikTok. The dashboard reads from the configured database and caches results briefly, so “live” should not be interpreted as instant streaming analytics.