BettaFish: AI-Powered Multi-Platform Public Opinion Monitoring in Minutes

Core Capabilities of BettaFish

Full‑stack data crawling : An AI‑driven crawler cluster runs 24/7, covering 30+ platforms (Weibo, Douyin, Xiaohongshu, Zhihu, Bilibili, foreign media, etc.). It captures posts, comments, images, and short videos automatically.

Multi‑Agent collaborative analysis : Three specialized agents (Query, Media, Insight) work together with a debate mechanism to reduce model bias. Sentiment classification reaches ~92% accuracy and extracts core user concerns.

Trend prediction & risk warning : Time‑series models trained on massive historical data forecast topic evolution for the next 24‑72 hours and flag high‑risk comments.

Automated report generation : The system aggregates data, creates charts (sentiment pie, trend line), extracts key insights, and produces a structured HTML report in about ten minutes.

Open‑source and customizable : All source code is publicly available, supports local deployment, can integrate with internal databases, and allows custom monitoring rules.

Typical Application Scenarios

1. Brand Public‑Opinion Monitoring

Workflow:

Enter keyword(s) and select platforms; set a monitoring window (e.g., 30 days).

The system automatically:

Crawls >1,200 items from selected platforms.

Analyzes sentiment (e.g., 65 % positive, 20 % negative, 15 % neutral).

Extracts top user demands (e.g., longer battery life, stable connection, more color options).

Generates an HTML report containing sentiment pie chart, trend line, demand list, high‑risk comment screenshots, and actionable recommendations.

2. Crisis Management with Early Warning

Real‑time monitoring of crisis‑related keywords (e.g., "brand quality issue") triggers alerts via DingTalk or email when high‑risk comments are detected. The system visualizes comment sources, predicts potential spread to hot‑search lists within six hours, and enables response within minutes.

3. Media & Self‑Media Topic Tracking

Input a topic keyword (e.g., "AI writing tools"). The system forecasts interest rise (e.g., +20 % over three days), helping content creators select high‑impact topics and monitor post‑publication sentiment.

Quick 3‑Step Setup for New Users

Step 1 – Deploy the System

Docker one‑click deployment (recommended) :

# Clone the repository
git clone https://github.com/666ghj/BettaFish.git
cd BettaFish
# Copy example environment file and edit as needed
cp .env.example .env
# Start services
docker compose up -d

Access the UI at http://localhost:5000.

Source‑code launch (for developers) :

# Install Python dependencies
pip install -r requirements.txt
# Install Playwright Chromium driver
playwright install chromium
# Configure .env (database and LLM parameters)
# Run the application
python app.py

Step 2 – Create a Monitoring Task

Define keywords (e.g., "Brand X", "AI tools").

Select platforms (Weibo, Douyin, Xiaohongshu, Zhihu, etc.).

Set monitoring mode (real‑time or custom time range).

Configure alert rules (e.g., trigger when negative sentiment >30 %).

Step 3 – View Results and Generate Report

The dashboard shows live counts of new content, sentiment ratios, and hot comments.

Click “Generate Report”; a structured HTML report is produced in ~10 minutes and can be downloaded.

Review the “Trend Prediction” module to anticipate topic evolution and plan proactive actions.

Technical Notes

BettaFish is designed to automate repetitive data collection, statistical calculation, and report formatting, allowing analysts to focus on strategic decisions. The community continuously adds features such as multilingual analysis, short‑video parsing, and enterprise data integration. Future work includes AI‑generated press‑release drafts.

Project repository: https://github.com/666ghj/BettaFish