TrendLens AI - Explainable AI for Social Media Trend Decline Prediction

Predict when trends collapse, understand why they decline, and act before it's too late.

TrendLens AI is an explainable Generative AI system that predicts when and why social media trends (memes, hashtags, viral topics) begin to lose momentum and eventually collapse. Unlike traditional trend analysis tools that focus only on growth, our system identifies early decline signals and provides clear, actionable explanations for each prediction.

The Problem

Social media trends have a lifecycle: they emerge, peak, decline, and eventually fade. Organizations waste millions on marketing campaigns for trends that are already declining, miss opportunities to exit before saturation, and struggle to understand why trends fail.

Key Challenges:

Reduced Engagement - Declining interaction rates signal trend fatigue
Influencer Disengagement - Key creators move on to new trends
Algorithmic Shifts - Platform algorithms deprioritize saturated content
Content Saturation - Market oversaturation leads to audience fatigue
Audience Fatigue - Users become desensitized to repetitive content

Business Impact:

Wasted marketing spend on declining trends
Missed opportunities to pivot strategies
Inability to proactively manage trend risks
Lack of understanding about trend lifecycle drivers

Our Solution

TrendLens AI combines multi-source data aggregation, explainable AI analysis, and actionable insights to provide a complete trend lifecycle prediction system.

Core Capabilities

Early Decline Detection - Identify decline signals before they become obvious
Explainable Predictions - Understand the "why" behind each prediction, not just the "what"
Multi-Platform Analysis - Track trends across Instagram, TikTok, Reddit, Twitter, Spotify, and more
Lifecycle Visualization - See where trends are in their lifecycle (Emerging → Rising → Peak → Declining → Fading)
Actionable Recommendations - Get specific guidance on when to jump in, proceed with caution, or avoid

How It Works

Data Collection - Continuously scrape and aggregate data from multiple social platforms
SMCE Scoring - Calculate trend health using Search Interest (S), Mention Velocity (M), Cultural Sentiment (C), and Engagement (E)
AI Analysis - Use Generative AI to analyze patterns and predict decline trajectories
Explainability - Generate human-readable explanations for each prediction
Visualization - Present insights through interactive dashboards and 3D network graphs

Features

1. Trend Lifecycle Predictions (`/trends`)

Track and predict the lifecycle of any social media trend in real-time.

Real-time Tracking - Monitor trends with live score updates
Lifecycle Stages - Visualize where trends are (Emerging, Rising, Peak, Declining, Fading)
Score Breakdown - See SMCE components (Search, Mentions, Culture, Engagement)
Trend Cards - Beautiful card-based interface showing trend health at a glance
Historical Analysis - View trend history with interactive sparkline charts

Key Metrics:

Current Score (0-100)
Percent Change (growth/decline rate)
SMCE Breakdown
Source distribution across platforms

2. Trend Analyzer (`/analyzer`)

Deep-dive analysis tool for understanding trend dynamics and decline factors.

Multi-Platform Sentiment - Analyze sentiment across TikTok, Instagram, Reddit, Twitter
Engagement Patterns - Identify engagement velocity changes
Cultural Context - Understand cultural relevance and mainstream adoption
Risk Factors - Identify potential decline drivers
Opportunity Detection - Find revival or pivot opportunities
Explainable Insights - Get AI-generated explanations for trend behavior

Analysis Components:

Platform-specific sentiment breakdown
Engagement velocity trends
Cultural vs. mainstream attention signals
Early warning signals
Brand impact assessment

3. Video Generator (`/video`)

AI-powered video content generation based on trending topics and lifecycle predictions.

Trend-Based Content - Generate videos aligned with trending topics
Lifecycle-Aware - Create content appropriate for trend stage
Multi-Format Support - Generate variations for different platforms
AI Captions - Auto-generate engaging captions
Performance Prediction - Estimate video performance before publishing

4. Campaign Generator (`/campaign`)

Create marketing campaigns optimized for trend lifecycle stages.

Trend Integration - Build campaigns around trending topics
Stage-Specific Strategies - Different approaches for Emerging vs. Peak trends
Multi-Week Planning - Plan campaigns across trend lifecycle
Content Suggestions - AI-generated content ideas
Risk Assessment - Evaluate campaign risks based on trend health

5. Campaign Orchestrator (`/orchestrate`)

Multi-agent orchestration system for complex, multi-step campaign execution.

What is Orchestration? A BFS (Breadth-First Search) based simulation engine that explores thousands of possible campaign paths, scores them, and recommends optimal strategies.

Key Features:

Path Simulation - Explore multiple campaign execution paths
AI-Powered Scoring - Evaluate paths using AI analysis
Pruning Engine - Automatically eliminate low-value paths
Multi-Agent Coordination - Coordinate multiple AI agents for different tasks
Session Management - Track orchestration sessions and results
Final Recommendations - Get top-scored campaign paths with explanations

Orchestration Process:

Initialize - Set up brand info, goals, and preferences
Generate Paths - Create multiple campaign execution paths
Score Nodes - Evaluate each path node using AI
Prune - Remove low-performing paths
Summarize - Generate AI summaries for top paths
Recommend - Present best paths with explanations

Use Cases:

Multi-platform campaign coordination
Complex influencer collaboration strategies
Long-term trend engagement planning
Risk-optimized campaign execution

6. Multi-Agent Social (`/multiagent`)

Advanced multi-agent system for complex social media analysis and decision-making.

Agent Coordination - Multiple specialized AI agents working together
Task Distribution - Distribute analysis tasks across agents
Collaborative Analysis - Agents share insights and refine predictions
Decision Support - Get multi-perspective recommendations

Architecture

Tech Stack

Frontend:

React 19 with Vite
Tailwind CSS + shadcn/ui components
Framer Motion for animations
React Three Fiber for 3D visualizations
Recharts for data visualization

Backend:

Express 5 (Node.js)
MongoDB with Mongoose
OpenAI GPT-4 for AI analysis
Node-cron for scheduled tasks
Puppeteer for web scraping

Data Sources:

Instagram (Reels, Posts)
TikTok
Reddit
Twitter/X
Spotify
Tavily (Web search)
Google Trends

Scrapers:

Instagram scraper (GraphQL + Magic Parameters)
TikTok scraper
Reddit API
Twitter API
Tavily API

Scoring System

Trends are scored using the SMCE Formula:

Score = 0.35×S + 0.30×M + 0.20×C + 0.15×E

Where:

S (Search Interest) - 35% weight - Google Trends, search volume
M (Mention Velocity) - 30% weight - Rate of mentions across platforms
C (Cultural Sentiment) - 20% weight - Cultural relevance and mainstream adoption
E (Engagement) - 15% weight - Likes, comments, shares, views

Score Interpretation:

80-100: Hot trend, peak engagement
60-79: Rising trend, growing momentum
40-59: Stable trend, moderate engagement
20-39: Declining trend, losing momentum
0-19: Fading trend, minimal engagement

Getting Started

Prerequisites

Node.js v20.12 or higher
MongoDB (local or cloud)
OpenAI API key
Instagram cookies (optional, for enhanced scraping)

Installation

Clone the repository

git clone git@github.com:Percobain/TrendLens.git
cd TrendLens

Install server dependencies

cd server
npm install

Install client dependencies

cd ../client
npm install

Configure environment variables

Server (server/.env):

MONGODB_URI=mongodb://localhost:27017/TrendLens
OPENAI_API_KEY=your_openai_api_key
PORT=3000

Client (client/.env):

VITE_API_URL=http://localhost:3000

Running the Application

Start MongoDB (if running locally)

mongod

Start the server

cd server
npm run dev

Start the client (in a new terminal)

cd client
npm run dev

Open the application

http://localhost:5173

Project Structure

datathong/
├── client/                 # React frontend
│   ├── src/
│   │   ├── components/    # React components
│   │   │   ├── dashboard/  # Dashboard components
│   │   │   ├── layout/     # Layout components
│   │   │   ├── multiagent/ # Multi-agent components
│   │   │   └── ui/         # UI components
│   │   ├── pages/          # Page components
│   │   ├── services/       # API services
│   │   └── utils/          # Utilities
│   └── package.json
│
├── server/                 # Express backend
│   ├── routes/             # API routes
│   ├── scrapers/           # Platform scrapers
│   ├── llm/                # AI analysis modules
│   ├── orchestrator/       # Campaign orchestration
│   ├── cron/               # Scheduled tasks
│   ├── db/                 # Database models
│   └── server.js           # Entry point
│
├── scraper/                # Instagram scraper
├── instacrapper/           # Instagram media scraper
└── commentScraper/         # Comment scraping tools

API Endpoints

Trends

GET /api/stocks - Get all tracked trends
POST /api/stocks/track - Start tracking a new trend
GET /api/stocks/:id - Get trend details
POST /api/stocks/refresh/:id - Refresh trend data
DELETE /api/stocks/:id - Stop tracking a trend

Analysis

POST /api/analyze - Analyze a trend
GET /api/discover - Discover trending topics

Campaign

POST /api/campaign/generate - Generate campaign
POST /api/orchestrate/start - Start orchestration
POST /api/orchestrate/step - Step through orchestration
GET /api/orchestrate/:sessionId - Get orchestration status

Video

POST /api/video/generate - Generate video content

Key Components

Dashboard Components

StockTicker - Live marquee showing trending topics
NetworkGraph - 3D interactive network visualization
TrendTable - Sortable table of trends
LifecycleGraph - Visual lifecycle stage indicator
DecisionCard - Actionable recommendations
ExplainabilityPanel - AI explanation viewer

Analysis Components

SentimentPanel - Platform-specific sentiment breakdown
TimeSeriesChart - Historical trend visualization
SourceGraph3D - 3D source network visualization

Data Flow

Collection - Scrapers collect data from social platforms
Storage - Data stored in MongoDB with timestamps
Scoring - SMCE scores calculated for each trend
Analysis - AI analyzes patterns and predicts decline
Explanation - AI generates human-readable explanations
Visualization - Frontend displays insights and predictions
Action - Users make decisions based on recommendations

Use Cases

For Marketers

Identify declining trends before wasting ad spend
Optimize campaign timing based on trend lifecycle
Understand why trends fail to adjust strategies
Discover emerging trends early

For Content Creators

Know when to jump on trends vs. when to avoid
Understand audience fatigue signals
Optimize content for trend stage
Plan content calendars around trend lifecycles

For Social Media Platforms

Monitor platform-wide trend health
Identify algorithmic shifts affecting trends
Understand user engagement patterns
Predict content saturation points

Future Enhancements

Real-time WebSocket updates
Advanced ML models for prediction accuracy
Integration with more platforms (YouTube, LinkedIn)
Automated campaign execution
Trend revival prediction
Competitive trend analysis
API for third-party integrations

License

This project is part of a hackathon submission. See individual component licenses for details.

Contributing

This is a hackathon project. For questions or contributions, please open an issue.

Contact

For questions about TrendLens AI, please refer to the project documentation or open an issue.

Built with 💖 by NGM GNG KRT BNG.

Name		Name	Last commit message	Last commit date
.vscode		.vscode
client		client
commentScraper		commentScraper
instacrapper		instacrapper
moltcot		moltcot
scraper		scraper
server		server
.DS_Store		.DS_Store
README.md		README.md

Latest commit

History

Folders and files

Repository files navigation