MicroForum

A microforum web application with role-based access, sentiment analysis, and an admin dashboard for analytics. Built with FastAPI, React, RabbitMQ, and Docker.

Table of Contents

• Overview
• Features
• Tech Stack
• Architecture
• Database Schema
• API Endpoints
• Sentiment Analysis
• Setup & Installation
• Usage
• Docker
• Contributing
• License

Overview

MicroForum is a modern forum application supporting user authentication, post and comment creation, sentiment analysis, and an admin dashboard for analytics. It is designed for easy deployment and scalability using Docker.

Features

• JWT-based authentication (signup/login)
• Role-based access (normal user/admin)
• Create, view, and delete posts and comments
• Sentiment analysis for posts and comments
• Admin dashboard with sentiment analytics (pie, line, bar charts)
• Asynchronous processing with RabbitMQ
• Containerized with Docker

Tech Stack

• Backend: FastAPI, SQLAlchemy, Pydantic, RabbitMQ
• Frontend: React.js, Tailwind CSS
• Database: SQLite
• AI/NLP: HuggingFace Transformers (Twitter-RoBERTa sentiment model)
• Containerization: Docker, Docker Compose

Architecture

• API Layer: Handles HTTP requests and responses (app/api/)
• Service Layer: Business logic and async jobs (app/services/)
• Repository Layer: Database operations (app/repositories/)
• Models: SQLAlchemy ORM models (app/models/)
• Schemas: Pydantic data validation (app/schemas/)
• Message Queue: RabbitMQ for async sentiment analysis

AI Model Used

This project uses a transformer-based sentiment analysis model from HuggingFace:

• Model: cardiffnlp/twitter-roberta-base-sentiment
• Type: RoBERTa, pre-trained and fine-tuned for sentiment analysis on social media and short text (Twitter).
• Why this model?
  • Outperformed other tested models (BERTweet, DistilBERT SST-2) in accuracy and speed for forum-style and social text.
  • Provides three sentiment classes: Positive, Neutral, Negative.
  • Easy integration with Python using the transformers library.

Message Queuing & Asynchronous Processing

• Message Queue: RabbitMQ is used to handle asynchronous tasks, specifically for sentiment analysis.
• How it works:
  • When a post or comment is created, a message is published to a queue.
  • A background worker consumes messages from the queue and performs sentiment analysis using the AI model.
  • The sentiment result is then stored back in the database.
• Benefits:
  • Decouples user-facing API from heavy AI processing.
  • Improves responsiveness and scalability.

General Backend Flow

1. Client sends a request (e.g., create post/comment) to the API Layer (FastAPI).
2. API Layer validates and forwards the request to the Service Layer.
3. Service Layer handles business logic and, if needed, publishes a message to RabbitMQ for async processing.
4. Repository Layer manages database operations (CRUD).
5. Worker listens to the message queue, processes sentiment analysis, and updates the database.
6. Client can fetch results (posts/comments with sentiment) via API endpoints.

Sentiment Analysis

Three transformer models were tested:

• BERTweet (finiteautomata/bertweet-base-sentiment-analysis)
• Twitter-RoBERTa (cardiffnlp/twitter-roberta-base-sentiment) ← Selected for best accuracy and speed
• DistilBERT SST-2 (distilbert-base-uncased-finetuned-sst-2-english)

Twitter-RoBERTa was chosen for its superior performance on social/forum text.

Setup & Installation

1. Clone the repository:

   git clone <repo-url>
   cd microforum-dev

2. Backend Setup:

   cd MicroForum/backend
   pip install -r requirements.txt
   # Set up environment variables as needed

3. Frontend Setup:

   cd ../frontend
   npm install

Usage

• Access the frontend at http://localhost:3000
• API available at http://localhost:8000/docs
• Admin dashboard available after login as admin

Contributing

1. Fork the repository
2. Create a new branch (git checkout -b feature/your-feature)
3. Commit your changes (git commit -am 'Add new feature')
4. Push to the branch (git push origin feature/your-feature)
5. Open a Pull Request

License

MIT License