CBMI - Cerberus-Misinformation

Overview

CBMI (Cerberus-Misinformation) is a Python-based tool designed to analyze social media activity on X and Reddit, helping you identify potential misinformation campaigns. It pulls posts based on your keywords, crunches the data with sentiment analysis, tracks user behavior, and cross-references it with Google Trends to give you a fuller picture. The output? A neat PDF report packed with insights—think sentiment trends, frequency charts, and a quick “is this fishy?” assessment—all backed by visuals you can actually use.

What It Does Under the Hood

Here’s the technical rundown of how CBMI works:

• Data Collection:

  • Hits X’s API (v2 via tweepy) and Reddit’s API (via praw) to grab posts matching your keywords within a specified time frame. It respects the free tier limits, capping at 10,000 posts total (split evenly between platforms).
  • Pulls Google Trends data (via pytrends) to see how your keywords are trending in search interest over the same period.

• Analysis:

  • Runs sentiment analysis on each post using TextBlob—scores range from -1 (negative) to 1 (positive)—and tracks daily averages to spot mood swings.
  • Counts post frequency per day with pandas to catch sudden spikes that might signal a coordinated push.
  • Extracts the top 10 keywords from posts using scikit-learn’s CountVectorizer for a peek at what’s buzzing.
  • Checks user behavior—flags accounts less than 30 days old posting more than 5 times as potential bots.

• Visualization:

  • Generates three plots with matplotlib (using the Agg backend for file-only output, handy for iOS or headless systems):
    • Post frequency over time.
    • Sentiment trend over time.
    • A dual-axis chart comparing social media posts to Google Trends interest.

• Misinformation Detection:

  • Looks for red flags like frequency spikes (3x the average), big sentiment shifts (>0.5), or bot-like activity. Correlates social media sentiment with Google Trends to see if chatter aligns with public curiosity.
  • Wraps it up in an “Intelligence Brief” with a verdict (e.g., “Potential misinformation detected”) and reasoning tied to the data.

• Reporting:

  • Compiles everything into a PDF via reportlab, including raw stats, plots, and the brief. You also get the raw .png plot files as a bonus.

It’s not rocket science, but it’s built to cut through the noise and give you something actionable—whether you’re tracking rumors, scams, or just curious about what’s trending.

Setup

Before you can use CBMI, you’ll need to set it up. Here’s the step-by-step:

Prerequisites

• Python: Version 3.9 or higher. Check with python --version.
• API Access: You’ll need keys for X and Reddit (details below).
• A Terminal: For CLI usage or initial setup.
• Optional (GUI): A desktop OS (Windows, Mac, Linux) with Tkinter installed for the GUI version.

Installation

1. Clone the Repository:

   • Grab the code from GitHub:

     git clone https://github.com/jdmx0/cbmi.git
     cd cbmi

2. Install Dependencies:

   • Run this in your terminal to get all the Python libraries CBMI needs:

     pip install -r requirements.txt

   • This pulls in tweepy, praw, textblob, pandas, matplotlib, reportlab, pytrends, and scikit-learn. On some systems, you might need pip install tk for the GUI.

3. Configure API Credentials:

   • Open cbmi/config.py in a text editor (e.g., Notepad, VS Code).
   • Add your API keys:
     • X: Get these from the X Developer Portal:
       • CONSUMER_KEY
       • CONSUMER_SECRET
       • ACCESS_TOKEN
       • ACCESS_TOKEN_SECRET
     • Reddit: Create an app at Reddit Apps:
       • REDDIT_CLIENT_ID
       • REDDIT_CLIENT_SECRET
       • Leave REDDIT_USER_AGENT as "cbmi" or tweak it if you want.
   • Save the file. These keys let CBMI talk to X and Reddit—without them, you’ll get nada.

Usage

CBMI offers two ways to run it: Command-Line Interface (CLI) for quick, no-frills analysis, and Graphical User Interface (GUI) for a point-and-click experience. Here’s how to use both.

CLI Usage

The CLI is perfect if you’re comfortable with a terminal and want to get straight to it.

How to Run

1. Navigate to the Folder:

   • In your terminal, move to the cbmi directory:

     cd path/to/cbmi

2. Launch the Script:

   • Type and hit Enter:

     python main.py

3. Answer the Prompts:

   • Keywords: Type space-separated words (e.g., election fraud). Hit Enter.
   • Sample Size: Enter a number between 1 and 1,000,000 (e.g., 500). It’s capped at 10,000 due to API limits. Hit Enter.
   • Start Time: Enter a date like 2023-10-01 (YYYY-MM-DD). Must be within X’s 7-day free API window. Hit Enter.
   • End Time: Enter a later date like 2023-10-07. Hit Enter.

4. Watch It Work:

   • The terminal shows progress: “Collecting up to 250 posts from X and Reddit...” then “PDF report generated: report.pdf”.
   • It grabs posts, analyzes them, and saves the report.

5. Check the Output:

   • Find report.pdf in the cbmi folder, along with .png plot files (frequency.png, sentiment.png, trends.png).
   • Open the PDF to see your analysis.

Example

$ python main.py
Enter keywords (space-separated): vaccine conspiracy
Enter sample size (1-1000000): 1000
Enter start time (YYYY-MM-DD): 2023-10-01
Enter end time (YYYY-MM-DD): 2023-10-07
Collecting up to 500 posts from X and Reddit...
PDF report generated: report.pdf