Page Chat

A book discovery and reading platform where users browse curated titles, read previews, and join safe book discussions.

Live Demo · Content Moderation · Tech Stack

---

Overview

Page Chat helps readers discover books by category, author, featured picks, and recommendations. Users can sign in with email or Google, read previews, track progress, and take part in book discussions.

Admins can manage books, authors, categories, members, uploaded files, and moderation from a custom dashboard.

Screenshots

Reading Experience	User Profile

Core Features

Feature	Description
Book discovery	Browse titles by category, author, featured picks, and recommendations
Reading experience	Read previews and track reading progress
Community discussions	Create discussions and comments around books
Authentication	Sign in with email/password or Google OAuth
Admin dashboard	Manage books, authors, categories, users, reports, and uploads
Object storage	Store book covers and PDFs with MinIO

Highlight: AI-Powered Content Moderation

The strongest part of this solution is its content safety pipeline. Page Chat protects discussions using a two-layer moderation system:

• Local filtering: catches profanity and suspicious patterns immediately using a custom blacklist
• Azure AI Content Safety: analyzes risky text for hate, violence, sexual, and self-harm categories

Depending on severity, content is either published, flagged for review, or hidden. Repeat offenders accumulate a violation score and are automatically suspended at 60 points.

Moderation Flow

flowchart TD
    A[User posts discussion or comment] --> B[before_validation: sanitize_profanity]
    B --> C{Local profanity detected?}
    C -->|No| D[Content saved as active]
    C -->|Yes| E[Mask offensive words in body]
    E --> F[SuspicionGate.monitor]
    F --> G[Increment user speech_violation_rating]
    G --> H[Azure Content Safety API analyzes original text]
    H --> I{Max severity}
    I -->|0-1| J[status: active]
    I -->|2-3| K[status: flagged + email user]
    I -->|4+| L[status: hidden + email user]
    G --> M{rating >= 60?}
    M -->|Yes| N[User auto-suspended]

Loading

Moderation Files

File	Purpose
back-end/app/models/concerns/profanity_filterable.rb	Entry point that runs before validation on discussions and comments
back-end/app/services/content_safety/suspicion_gate.rb	Local suspicion rules and Azure trigger
back-end/app/services/content_safety/azure_content_safety.rb	Azure AI Content Safety API client
back-end/config/blacklist.yml	Custom profanity blacklist

Tech Stack

Layer	Stack
Frontend	React, Vite, Redux Toolkit, Tailwind CSS
Backend	Ruby on Rails 7, Devise JWT, OmniAuth (Google)
Storage	PostgreSQL, Redis, MinIO (S3-compatible)
Moderation	Obscenity gem + Azure AI Content Safety
Runtime	Docker Compose, Nginx, Let's Encrypt SSL

Project Structure

page_chat/
├── back-end/     # Rails API + Trestle admin
├── front-end/    # React SPA
└── docker-compose.yml

Local Development

Development uses the production PostgreSQL on the server via an SSH tunnel.

Open the database tunnel:

ssh -L 5432:localhost:5432 epic@89.167.4.31 -N -f

Start the backend:

cd back-end && rails s

Close the tunnel:

pkill -f "ssh -L 5432"

---

Page Chat — safer book discussions powered by Rails, React, and Azure AI.