Bot Detection System

Overview

The bot detection system is a comprehensive multi-layered approach to identify automated user behavior and suspicious activity patterns on the confession platform. It uses advanced algorithms to analyze user behavior and maintain platform integrity.

🎯 System Architecture

Core Components

BotDetectionService (src/lib/bot-detection.ts)
- Main service class with singleton pattern
- Comprehensive bot detection algorithms
- Risk level classification
- Statistical analysis
Database Integration (src/lib/db-utils.ts)
- Enhanced reputation calculation with bot penalties
- Real-time activity monitoring
- Bot probability updates
- Historical tracking
Admin Dashboard (src/components/BotDetectionPanel.tsx)
- Real-time bot detection statistics
- Risk level visualization
- Detection method explanations
- User management interface

🔍 Detection Methods

1. Activity Pattern Analysis

Excessive Activity Detection: Monitors daily action counts (>50 actions = high risk)
Activity Burst Analysis: Identifies unnatural activity spikes
Consistent Pattern Recognition: Detects unnaturally regular behavior patterns

2. Content Analysis

Repetitive Content Detection: Analyzes content similarity (>70% similarity = high risk)
Generic/Bot-like Content: Identifies templated or automated content
Content Length Analysis: Detects suspicious length patterns

3. Timing Analysis

Regular Interval Detection: Identifies automated posting schedules
24/7 Activity Monitoring: Detects continuous activity patterns
Timezone Inconsistency Checks: Identifies multi-timezone bot networks

4. Engagement Analysis

Low Engagement Despite High Activity: Detects bots that post but don't interact
One-sided Interaction Detection: Identifies bots that only post or only vote
Social Graph Analysis: Analyzes user interaction patterns

5. Device/IP Analysis

Multiple Account Detection: Identifies accounts from same source
Device Fingerprinting: Tracks device patterns (future implementation)
IP Address Analysis: Monitors IP-based patterns

6. Social Graph Analysis

Isolation Score: Measures social connection patterns
Artificial Social Patterns: Detects fake social interactions
Network Analysis: Identifies bot networks

📊 Risk Classification

Risk Levels

Level	Probability	Action Required	Description
CRITICAL	80%+	Immediate suspension	High confidence bot detection
HIGH	60-79%	Enhanced monitoring	Strong bot indicators
MEDIUM	30-59%	Regular monitoring	Moderate suspicion
LOW	0-29%	Normal monitoring	Normal activity patterns

Bot Probability Calculation

Total Score = Activity Score + Content Score + Timing Score + 
              Engagement Score + Device Score + Social Score

Bot Probability = (Total Score / Max Score) * 100

🛠️ Implementation Details

Database Schema Updates

model User {
  // Bot detection fields
  botProbability     Int @default(0) // 0-100, higher = more likely bot
  isFlagged          Boolean @default(false)
  verificationLevel   String @default("NEW_USER")
  
  // Activity tracking
  dailyConfessions   Int @default(0)
  dailyVotes         Int @default(0)
  dailyComments      Int @default(0)
  confessionsCreated Int @default(0)
  votesCast          Int @default(0)
  commentsCreated    Int @default(0)
}

model ReputationHistory {
  // Enhanced tracking
  botProbability     Int?
  botIndicators      String[] // Array of bot detection indicators
  riskLevel          String? // LOW, MEDIUM, HIGH, CRITICAL
}

API Endpoints

GET /api/admin/bot-detection/stats
- Returns bot detection statistics
- Total users, suspicious users, critical users, average probability
GET /api/admin/bot-detection/[userId]
- Returns detailed bot detection info for specific user
- Probability, indicators, risk level, recommendations

Integration Points

User Activity Monitoring

// Called on every user action
await monitorUserActivity(userId, 'confession' | 'vote' | 'comment');

Reputation Calculation

// Bot penalty applied to reputation
const botPenalty = botProbability > 50 ? (botProbability - 50) * 0.5 : 0;
const reputationScore = activityScore + qualityScore + trustScore - botPenalty;

Real-time Flagging

// Automatic flagging for critical risk
if (botDetectionResult.riskLevel === 'CRITICAL') {
  await flagUser(userId);
}

📈 Monitoring & Analytics

Key Metrics

Bot Detection Rate: Percentage of users flagged as suspicious
Critical Risk Users: Number of users with 80%+ bot probability
Average Bot Probability: Mean bot probability across all users
Detection Accuracy: False positive/negative rates

Dashboard Features

Real-time Statistics: Live updates of bot detection metrics
Risk Level Visualization: Color-coded risk indicators
Detection Method Breakdown: Explanation of detection algorithms
User Management: Flag/unflag users with detailed reasoning

🔧 Configuration

Detection Thresholds

const DETECTION_THRESHOLDS = {
  EXCESSIVE_ACTIVITY: 50,        // Daily actions
  HIGH_ACTIVITY: 30,             // Daily actions
  REPETITIVE_CONTENT: 0.7,       // Similarity threshold
  GENERIC_CONTENT: 0.6,          // Generic content threshold
  REGULAR_INTERVALS: 0.8,        // Timing regularity
  CONTINUOUS_ACTIVITY: true,     // 24/7 activity flag
  LOW_ENGAGEMENT_RATIO: 0.1,     // Engagement threshold
  ONE_SIDED_INTERACTIONS: 0.8,   // Interaction pattern threshold
};

Risk Level Thresholds

const RISK_THRESHOLDS = {
  CRITICAL: 80,  // 80%+ bot probability
  HIGH: 60,      // 60-79% bot probability
  MEDIUM: 30,    // 30-59% bot probability
  LOW: 0,        // 0-29% bot probability
};

🚀 Usage Examples

Basic Bot Detection

import { botDetectionService } from '@/lib/bot-detection';

// Detect bot for a user
const result = await botDetectionService.detectBot(userId);
console.log(`Bot probability: ${result.probability}%`);
console.log(`Risk level: ${result.riskLevel}`);
console.log(`Indicators: ${result.indicators.join(', ')}`);

Admin Dashboard Integration

import { BotDetectionPanel } from '@/components/BotDetectionPanel';

// Add to admin page
<BotDetectionPanel />

Real-time Monitoring

// Monitor user activity
await monitorUserActivity(userId, 'confession');

// Get bot detection stats
const stats = await getBotDetectionStats();
console.log(`Critical users: ${stats.criticalUsers}`);

🔒 Security Considerations

Privacy Protection

Bot detection data is stored securely
Personal information is not exposed in bot analysis
Detection algorithms focus on behavior patterns, not personal data

False Positive Mitigation

Multiple detection methods reduce false positives
Manual review process for flagged users
Appeal system for incorrectly flagged accounts

Performance Optimization

Efficient database queries for large user bases
Caching of detection results
Asynchronous processing for real-time monitoring

📋 Future Enhancements

Planned Features

Machine Learning Integration: Advanced ML models for pattern recognition
Behavioral Biometrics: Mouse movement and typing pattern analysis
Network Analysis: Detection of coordinated bot networks
Real-time Alerts: Instant notifications for critical detections
Automated Actions: Automatic suspension for confirmed bots

Advanced Detection Methods

Natural Language Processing: Content sentiment and style analysis
Image Analysis: Profile picture and content image analysis
Geographic Analysis: Location-based pattern detection
Temporal Analysis: Time-based behavior pattern recognition

🎯 Benefits

For Platform Integrity

Reduced Spam: Automated detection prevents spam content
Quality Control: Maintains high-quality user interactions
Trust Building: Creates a safer environment for users
Resource Protection: Prevents abuse of platform resources

For User Experience

Cleaner Feed: Less spam and automated content
Better Engagement: Higher quality interactions
Fair Competition: Prevents bot-driven unfair advantages
Community Health: Maintains authentic community interactions

For Administrators

Automated Monitoring: Reduces manual review workload
Detailed Analytics: Comprehensive bot detection statistics
Proactive Management: Early detection of suspicious activity
Scalable Solution: Handles large user bases efficiently

🔄 Maintenance

Regular Updates

Monitor detection accuracy rates
Adjust thresholds based on false positive/negative rates
Update detection algorithms based on new bot patterns
Review and refine risk classification system

Performance Monitoring

Track detection processing times
Monitor database query performance
Optimize algorithms for large-scale deployment
Scale infrastructure as user base grows

This bot detection system provides comprehensive protection against automated abuse while maintaining user privacy and platform performance.

FilesExpand file tree

BOT_DETECTION_SYSTEM.md

Latest commit

History