Cross-Platform Sentiment Analysis & Malicious Post Detection

A system for analyzing public opinion and detecting malicious posts across Reddit and Twitter.
Built with Python, NLP models, PostgreSQL on AWS, and Plotly Dash.

---

🚀 Overview

This project helps brands monitor their online reputation by:

• Performing sentiment analysis on Reddit posts and comments.
• Detecting malicious or harmful posts (hate speech, fake news, toxic content) on Twitter.
• Providing real-time insights through an interactive dashboard.

---

🏗️ How It Works

1. Data Ingestion

   • Reddit: Streamed using PRAW.
   • Twitter: CSV uploads.

2. Analysis

   • Models: RoBERTa, DistilBERT, VADER.
   • Weighted scoring system to classify sentiment and detect malice.

3. Storage & Visualization

   • Data stored in PostgreSQL (AWS RDS).
   • Interactive dashboard built with Plotly Dash.

---

📊 Features

• Real-time Reddit streaming & Twitter CSV ingestion.
• Sentiment classification (positive, neutral, negative).
• Malicious post detection (flagging toxic content).
• Dashboard with pie charts, line graphs, and filterable tables.

---

🔧 Tech Stack

• Python
• NLP Models: RoBERTa, DistilBERT, VADER
• Database: PostgreSQL (AWS RDS)
• Dashboard: Plotly Dash, Flask
• Deployment: AWS EC2

---

📌 Future Work

• Train on dedicated malicious post datasets.
• Add API integrations for real-time Twitter ingestion.
• Expand to more social platforms (Instagram, YouTube, etc.).

---

👩‍💻 Contributors

• Dhananjay Surti
• Isha Das
• Ben Flock
• Zach Youssef

---