Home

PATAS Core Documentation

Pattern-Adaptive Transmodal Anti-Spam System

PATAS Core technical documentation. This wiki contains technical documentation for integration and deployment.

---

Quick Navigation

Getting Started

• Quick Start Guide - Installation and basic usage
• Architecture - System architecture and design principles
• Code Overview - Code structure and navigation guide
• Configuration - Configuration reference and environment variables
• Scaling & Cost Design - How PATAS handles millions of logs efficiently
• Comparison and Positioning - PATAS vs traditional approaches

Two-Stage Processing & Scaling

• Scaling & Cost Design ⭐ - One-pager on how PATAS scales efficiently
• Two-Stage Analysis - Detailed analysis of the two-stage approach
• Two-Stage Tuning Guide - Configuration and threshold tuning
• Pattern Mining Scaling - Technical analysis of scaling pattern mining
• Threshold Tuning Results - Real-world tuning results

Production & Benchmarks

• Production Benchmark Analysis - Detailed benchmark results
• Benchmark Report (Real Data) - Real Telegram data benchmark
• Benchmark Summary for Pitch - Concise metrics for presentations
• Final Production Report - Complete production readiness report

Implementation & Phases

• Phase 1 Complete - Phase 1 improvements summary
• Phase 2 Complete - Phase 2 implementation summary
• Two-Stage Implementation Summary - Complete implementation overview
• Implementation Complete - Final implementation report
• Testing Report (Two-Stage) - Comprehensive testing results

Core Concepts

• Safety Profiles - Conservative, Balanced, and Aggressive profiles
• Architecture - System architecture and design principles

Telegram Integration

• Engineering Notes for Telegram - Technical overview for Telegram engineers
• Telegram Safety Guide - Safety profiles and enforcement model

API Documentation

• API Reference - Complete API endpoint documentation
• API Quickstart - Quick start guide for API usage

Safety and Privacy

• Privacy and Data Protection - Privacy modes and data handling
• LLM Usage - LLM integration and privacy guarantees
• Security - Security considerations and best practices

---

What is PATAS?

PATAS is an autonomous pattern discovery and rule management system for anti-spam operations. It analyzes historical message logs, automatically discovers spam patterns, generates safe blocking rules, and evaluates their effectiveness before deployment.

Key Characteristics:

• Signal engine, not enforcement - PATAS provides patterns and metrics that inform anti-spam decisions
• On-premise deployment - Designed for deployment within your infrastructure
• Two-stage processing - Fast scanning + deep analysis for 70-90% cost reduction
• Deterministic and rule-based - Core engine is deterministic; ML/LLM is optional
• Safety-first design - Multiple safety profiles with clear risk boundaries

---

Core Workflow

1. Ingest - Load historical message logs into PATAS
2. Discover - Automatically identify recurring spam patterns (two-stage: fast scan + deep analysis)
3. Generate - Create safe SQL rules from discovered patterns
4. Evaluate - Test rules on historical data (shadow mode)
5. Promote - Activate rules that meet safety thresholds
6. Monitor - Track rule performance and deprecate underperforming rules

---

Documentation Structure

Architecture and Design:

• Start with Architecture for high-level understanding
• Review Code Overview to understand codebase structure
• See Configuration for setup and tuning

Scaling and Performance:

• Scaling & Cost Design - How PATAS scales efficiently
• Two-Stage Analysis - Two-stage processing deep dive
• Production Benchmark Analysis - Real-world performance metrics

Telegram Integration:

• Start with Engineering Notes for Telegram
• Review Telegram Safety Guide

API and Operations:

• Review API Reference for integration
• See Configuration and Quick Start for deployment

---

Repository

• GitHub: kiku-jw/PATAS