Transcript Analyzer

A comprehensive web application for analyzing consumer-agent conversation transcripts with AI-powered insights including sentiment analysis, intent classification, predictive analytics, and automated recommendations.

Features

📊 Comprehensive Analysis

Traditional Analysis:

Sentiment Analysis: Overall and per-message sentiment scoring (0-100 scale)
Topic Extraction: Automatic identification of key topics using TF-IDF
Keyword Extraction: Most frequent and relevant terms
Performance Metrics: Message counts, response times, conversation duration
Agent Performance Scoring: Professionalism and effectiveness metrics
Customer Satisfaction Estimation: Based on language patterns and indicators

AI-Powered Analysis:

Summarization: 2-3 sentence summaries with key points and action items
Advanced Sentiment: Emotion detection, sentiment trajectory, turning points, empathy scoring
Intent Classification: Primary and secondary intents with complexity levels
Agent Performance: Communication quality, problem-solving, compliance, personalization
QA Insights: Policy violations, script adherence, risk flags, best practices
Customer Journey: Churn risk prediction, personality profiling, LTV indicators

🔮 Predictive Insights

Sentiment Trend Analysis: Historical trends with forecasting and anomaly detection
Churn Risk Prediction: ML-based customer churn risk scoring with risk factors
Intent Impact Matrix: Volume vs Sentiment visualization with complexity mapping
AI-Powered Recommendations: Automated priority-based action recommendations

🔗 LivePerson Integration

Multi-Account Support: Manage multiple LivePerson accounts
OAuth 1.0 Authentication: Secure API integration
Conversation Fetching: Bulk import with date range and skill filtering
Real-time Progress: Live fetch status with conversation counts

🎨 Premium Web Interface

Interactive Dashboard: Real-time statistics and visualizations
Interactive Explorer: 4 visualization panels (Intent Impact, Empathy, Churn Risk, Resolution)
AI Insights Panel: Automated recommendations with priority badges
Multi-Account Filtering: Filter all views by LivePerson account
Drag & Drop Upload: Easy file upload with progress tracking
Advanced Filtering: Filter results by sentiment, intent, complexity, date
Export Functionality: Download results in CSV or JSON format
Responsive Design: Works beautifully on all devices

🚀 Technology Stack

Backend: Node.js, Express, SQLite
Frontend: React 18, Vite, Framer Motion
Analysis: Natural (NLP), Sentiment Analysis
AI: OpenAI GPT-4, Google Gemini (configurable)
Visualization: Recharts, Chart.js
Design: Modern glassmorphism with vibrant gradients

Installation

Prerequisites

Node.js 18+ installed
npm or yarn package manager
OpenAI API key or Google Gemini API key (for AI analysis)

Setup

Clone or navigate to the project directory

cd transcript-analyzer

Install backend dependencies

npm install

Install frontend dependencies

cd client
npm install
cd ..

Configure AI Settings (Optional) Create a .env file in the root directory:

OPENAI_API_KEY=your_openai_api_key
# OR
GOOGLE_API_KEY=your_gemini_api_key

Run Database Migrations (if upgrading from older version)

node scripts/run-migrations.js

Running the Application

Start the Backend Server

npm run dev

The API server will start on http://localhost:3000

Start the Frontend (in a new terminal)

cd client
npm run dev

The web app will be available at http://localhost:5173

Usage

1. Import Conversations

Option A: Upload Files Navigate to the Import tab → Upload section and upload conversation transcripts in JSON or text format.

Option B: LivePerson Integration

Go to Import tab → LivePerson section
Add your LivePerson account credentials (OAuth 1.0)
Select date range and optional skills filter
Click "Fetch Conversations" to import

JSON Format:

[
  {
    "conversation_id": "conv_001",
    "conversation_date": "2025-11-26",
    "transcript_details": "Agent: Hello, how can I help you today?\nCustomer: I have a question about my order..."
  }
]

2. Run Analysis

Go to the Analysis tab:

Traditional Analysis: Click "Run Traditional Analysis" for fast rule-based analysis
AI Analysis: Click "Run AI Analysis" for deep AI-powered insights
View results in the combined table with all metrics

3. View Dashboard

The Dashboard provides an overview with:

Total conversations and analysis counts
Average sentiment, CSAT, and agent scores
Sentiment distribution charts
Recent conversations table
Sentiment Trend Chart: Historical trends with forecast and anomalies
Churn Risk Panel: High-risk conversation identification

4. Explore Insights

The Explore tab features:

AI-Powered Insights Panel: Top 3 automated recommendations
Intent Impact Analysis: Volume vs Sentiment bubble chart
Empathy Distribution: Agent empathy score distribution
Churn Risk Visuals: Risk level breakdown
Resolution Status: Resolved vs unresolved issues

5. Export Results

Use the Export button in Analysis tabs to download results in:

JSON: Structured data for programmatic access
CSV: Spreadsheet-compatible for Excel/Google Sheets

API Documentation

Conversations Endpoints

POST /api/conversations/upload - Upload file with conversations
POST /api/conversations/bulk - Bulk upload via JSON
GET /api/conversations - List all conversations (paginated)
GET /api/conversations/:id - Get specific conversation
DELETE /api/conversations/:id - Delete conversation

LivePerson Endpoints

GET /api/liveperson/accounts - List all LivePerson accounts
POST /api/liveperson/accounts - Create new account
PUT /api/liveperson/accounts/:id - Update account
DELETE /api/liveperson/accounts/:id - Delete account
POST /api/liveperson/fetch - Fetch conversations from LivePerson
GET /api/liveperson/export-csv - Export conversations to CSV

Analysis Endpoints

POST /api/analysis/run - Run traditional analysis
GET /api/analysis/results - Get analysis results (with filters)
GET /api/analysis/dashboard - Get dashboard metrics
GET /api/analysis/export?format=json|csv - Export results

AI Analysis Endpoints

POST /api/ai-analysis/run - Run AI analysis on conversations
GET /api/ai-analysis/results - Get AI analysis results (paginated, with filters)
GET /api/ai-analysis/summary/:id - Get detailed AI analysis for a conversation
GET /api/ai-analysis/insights - Get aggregated AI insights
GET /api/ai-analysis/costs - Get cost tracking data
GET /api/ai-analysis/stats - Get dashboard statistics
GET /api/ai-analysis/trends - Get sentiment trends and forecast
GET /api/ai-analysis/churn-risks - Get churn risk conversations
POST /api/ai-analysis/calculate-churn - Calculate churn risk scores
GET /api/ai-analysis/intents - Get aggregated intent statistics
GET /api/ai-analysis/intent-insights - Get AI-powered intent recommendations

Configuration Endpoints

GET /api/analysis-config - Get all analysis configurations
PUT /api/analysis-config - Update analysis configuration
GET /api/metric-configs - Get all metric configurations
POST /api/metric-configs - Create or update metric config
GET /api/prompts - Get all AI prompts
POST /api/prompts - Create new prompt
GET /api/settings - Get AI settings
PUT /api/settings - Update AI settings

Architecture

transcript-analyzer/
├── server.js                    # Express server
├── database.js                  # SQLite database setup
├── analyzer.js                  # Traditional analysis engine
├── aiAnalyzer.js               # AI analysis engine
├── llmService.js               # LLM provider abstraction
├── routes/
│   ├── conversations.js        # Conversation endpoints
│   ├── analysis.js             # Traditional analysis endpoints
│   ├── aiAnalysis.js           # AI analysis endpoints
│   ├── liveperson.js           # LivePerson integration
│   └── ...                     # Config endpoints
├── services/
│   ├── livepersonService.js    # LivePerson API client
│   ├── trendAnalysis.js        # Sentiment trend forecasting
│   ├── churnPrediction.js      # Churn risk ML model
│   └── intentInsights.js       # AI-powered recommendations
├── migrations/
│   └── add_predictive_insights.js
├── scripts/
│   ├── run-migrations.js       # Migration runner
│   ├── backfill_trends.js      # Trend data backfill
│   └── README.md               # Scripts documentation
├── client/
│   ├── src/
│   │   ├── App.jsx             # Main app component
│   │   ├── components/         # React components
│   │   │   ├── Dashboard.jsx
│   │   │   ├── InteractiveExplorer.jsx
│   │   │   ├── LivePersonFetch.jsx
│   │   │   └── visualizations/
│   │   │       ├── IntentAnalysisChart.jsx
│   │   │       ├── IntentInsightsPanel.jsx
│   │   │       ├── SentimentTrendChart.jsx
│   │   │       └── ChurnRiskPanel.jsx
│   │   ├── context/
│   │   │   └── AccountContext.jsx
│   │   ├── utils/
│   │   │   └── api.js          # API client
│   │   └── index.css           # Design system
│   └── index.html
├── DATABASE.md                  # Database schema documentation
└── package.json

Performance

Handles unlimited conversations with server-side aggregation
Traditional analysis: ~100 conversations/second
AI analysis: ~5-10 conversations/minute (depends on LLM provider)
Efficient SQLite database with indexes for fast queries
Optimized frontend with lazy loading and code splitting

Database

The application uses SQLite with 11 tables:

conversations - Conversation transcripts
liveperson_accounts - LivePerson account credentials
analysis_results - Traditional analysis results
ai_analysis_results - AI analysis results with 30+ metrics
sentiment_trends - Daily aggregated sentiment trends
ai_cost_tracking - AI usage cost tracking
ai_settings - AI configuration
ai_prompts - Custom AI prompt templates
metrics - General metrics storage
metric_configs - Custom metric configurations
analysis_config - Analysis algorithm configuration

See DATABASE.md for complete schema documentation.

Troubleshooting

Backend won't start:

Ensure port 3000 is available
Check Node.js version (18+)
Verify all dependencies are installed
Check for database migration errors

Frontend can't connect to backend:

Ensure backend is running on port 3000
Check CORS settings in server.js
Verify API_BASE_URL in client/src/utils/api.js

AI Analysis not working:

Verify API key is set in Settings tab or .env file
Check AI provider selection (OpenAI vs Gemini)
Monitor cost tracking to ensure budget limits aren't exceeded
View browser console for error messages

LivePerson fetch failing:

Verify OAuth credentials are correct
Check account ID matches your LivePerson account
Ensure date range is valid
Check network connectivity to LivePerson API

License

MIT

Support

For issues or questions:

Check browser console for error messages
Review server logs for backend errors
Verify database schema with sqlite3 transcripts.db .schema
Check DATABASE.md for schema documentation
Review scripts/README.md for migration help

Name		Name	Last commit message	Last commit date
Latest commit History 30 Commits
client		client
migrations		migrations
routes		routes
scripts		scripts
services		services
.env.example		.env.example
.gitignore		.gitignore
API.md		API.md
DATABASE.md		DATABASE.md
README.md		README.md
aiAnalyzer.js		aiAnalyzer.js
analyzer.js		analyzer.js
database.js		database.js
database_schema.sql		database_schema.sql
fetch_transcript		fetch_transcript
llmService.js		llmService.js
migrate-liveperson-endpoint.js		migrate-liveperson-endpoint.js
migrate-liveperson-settings.js		migrate-liveperson-settings.js
migrate-liveperson.js		migrate-liveperson.js
package-lock.json		package-lock.json
package.json		package.json
prompts.js		prompts.js
sample_conversations.json		sample_conversations.json
schema.sql		schema.sql
server.js		server.js
test_dates.js		test_dates.js
test_topics.js		test_topics.js

jeffreyha-dev/transcript-analyzer

Folders and files

Latest commit

History

Repository files navigation