Anonalyze: An NLP-Enhanced and ML-Driven Platform for Sentiment and Insight Extraction

preview.mp4

Anonalyze is a research-driven AI, ML, and NLP-powered web platform developed as part of an undergraduate thesis project. The platform focuses on analyzing anonymous user-generated content through Machine Learning (ML) and Natural Language Processing (NLP) techniques.

• Sentiment Analysis (positive, negative, neutral)
• Emotion Classification (joy, sadness, anger, fear, love, surprise)
• Keyword & Keyphrase Extraction (RAKE algorithm)
• Discussion Summarization (LLM-based summaries)

Model Training Process

This project utilizes a sentiment and emotion classification model using Machine Learning (ML) and Natural Language Processing (NLP). The system analyzes text and predicts:

• Sentiment: positive, negative, or neutral
• Emotion: sadness, joy, love, anger, fear, or surprise

The core model utilizes a Multinomial Naive Bayes classifier, trained on a large dataset of Twitter messages.

Model Training Process:

• Repository: https://github.com/hyoaru/anonalyze-process
• See the training process live

Tech Stack

• Frontend: React + Typescript + Tanstack
• Backend API: Laravel (PHP) + Flask (Python)
• Machine Learning: Python (scikit-learn, NLTK, Pandas, NumPy)
• Containerization: Docker & Docker Compose
• Database: MySQL

Setup Guide

Follow the steps below to run the project in development mode:

1. Clone the Repository

git clone https://github.com/hyoaru/anonalyze.git
cd anonalyze

2. Environment Setup

cat .env.example > .env
cat ./client/web/.env.example > .env
cat ./api/core/.env.example > .env   # Make sure to populate GEMINI_API key
cat ./api/ml/.env.example > .env

3. Run Docker Compose

sudo docker compose -f docker-compose.development.yaml up

4. Configure Core API

Enter the container:

docker exec -it anonalyze_api_core bash

Run the following commands inside:

php artisan key:generate
php artisan migrate --database=mysql
php artisan db:seed

Ports and Access

Service	URL / Hostname	Description
Web Client (React)	http://localhost:8000	Main web interface
Core API (Laravel)	http://localhost:8002	Backend API for application logic
ML API (Flask)	http://localhost:8003	Machine learning microservice (sentiment, emotion, keywords, summaries)
Database Admin (phpMyAdmin)	http://localhost:7999	MySQL database administration
SMTP Mailhog	http://localhost:7998	Local mail testing (emails capture)

Project Structure

.
├── api
│ ├── core-api # Laravel backend for core application logic and REST API
│ │ ├── app/ # Controllers, Models, Services, Repositories, Policies
│ │ ├── database/ # Migrations, Seeders, Factories
│ │ ├── routes/ # API and web routes
│ │ ├── config/ # Laravel configurations
│ │ ├── tests/ # Unit and feature tests
│ │ ├── artisan # Laravel CLI
│ │ └── public/ # Public assets and index.php
│ │
│ └── ml-api # Python ML microservice
│ ├── app/
│ │ ├── api/ # API namespaces (emotion, sentiment, concept, summary)
│ │ ├── controllers/ # REST controllers for ML endpoints
│ │ ├── configurations/ # Model and feature selector configs
│ │ ├── models/ # ML models (Naive Bayes, RAKE, TF-IDF, Chi-squared)
│ │ ├── utils/ # Preprocessing utilities (denoising, stopwords, lemmatization)
│ │ └── data/pkl/ # Serialized models, vectorizers, and selectors
│ ├── main_development.py
│ ├── main_production.py
│ └── requirements.txt
│
├── client
│ └── web # React frontend
│ ├── public/ # Public assets
│ ├── src/ # Components, pages, and logic
│ ├── index.html
│ ├── package.json
│ └── Dockerfile
│
├── docker-compose.development.yaml
├── docker-compose.production.yaml
├── .env.example
└── README.md

Features

• Anonymous Posting – Encourages open communication without fear of exposure.
• ML-Driven Sentiment Analysis – Classifies posts as positive, neutral, or negative.
• Emotion Classification – Identifies six core emotions: joy, sadness, anger, fear, love, surprise.
• Keyword Extraction – Highlights important discussion terms using the RAKE algorithm.
• Thread Summarization – Uses LLMs to condense long threads into key takeaways.
• Dashboard & Visualization – Real-time insights and analytics.

Academic Context

This project was developed as an undergraduate thesis to demonstrate the application of NLP and ML in processing user-generated content. It aims to contribute to research in machine learning, sentiment analysis, emotion detection, and automated summarization, while also showcasing a practical implementation through a web-based system.