Skip to content
BranchesTags

Repository files navigation

SENTIMENT & URGENCY DETECTOR

🧠 AI-Powered Customer Support Ticket Intelligence System

Sentiment & Urgency Detector is an AI-based system that analyzes customer support tickets to automatically detect Sentiment, Urgency, and Churn Risk using Large Language Models (OpenAI + Hugging Face) and triggers real-time alerts through Discord for critical issues.

📌 Submission Identity (Placement Drive)

Field Details
🏢 Organization Infinity Computer Solutions
📂 Category Service Desk
👥 Team Number Team 8
🆔 UC ID UC-4
🚀 Project Title Sentiment & Urgency Detector
🧠 Domain AI / LLM-Based Automation System
📅 Type Placement Drive Submission

👨‍💻 Team Members

Name Role
Leela Project Lead & System Design
Mohan Backend Developer (FastAPI Integration)
Lalitha AI/ML Engineer (OpenAI + Hugging Face)
Sunitha Testing, Documentation & UI Support

🌟 Project Overview

Customer support teams receive thousands of tickets daily. Many critical messages containing anger, urgency, frustration, or cancellation intent are often missed due to manual handling.

This results in:

  • Delayed responses
  • Poor customer satisfaction
  • Increased customer churn
  • Revenue loss

💡 Our Solution

The system automatically: ✔ Detects sentiment of customer messages ✔ Calculates urgency level ✔ Identifies churn-risk behavior ✔ Generates AI-based explanations ✔ Sends real-time alerts via Discord

❗ Problem Statement

Organizations face difficulty in identifying high-priority tickets among large volumes of incoming customer requests.

Manual handling leads to:

  • Slow response time
  • Missed urgent complaints
  • Inefficient escalation process

This project solves it using AI-driven automation and LLM intelligence.

🚀 Live Application

🌐 https://sentiment-and-urgency-detector-dqg64ccmtf28dul7pead9m.streamlit.app/

🎥 Live Demo Video

📽️ https://drive.google.com/file/d/1l609UCGY_uJ-B57SB1o6_vdm3HbB9MO6/view?usp=sharing

✨ Key Features

🤖 AI Intelligence Features

  • 😊 Sentiment Detection (Positive / Neutral / Negative)
  • ⚡ Urgency Scoring System
  • 🚨 Churn Risk Prediction
  • 🧠 Explainable AI Reasoning
  • 📊 Confidence Score Generation

🔔 Automation Features

  • Discord Webhook Alerts
  • Real-time Ticket Analysis
  • Automatic Escalation Trigger

📦 System Features

  • REST API (FastAPI)
  • JSON Structured Output
  • Streamlit UI Support

🏗️ System Architecture

Customer Ticket Input
        │
        ▼
LLM Processing Layer (OpenAI / Hugging Face)
        │
        ├── Sentiment Analysis
        ├── Urgency Detection
        ├── Churn Risk Analysis
        │
        ▼
Risk Evaluation Engine
        │
   High Risk?
        │
     YES ▼
Discord Webhook Alert System
        │
Support Team Notification

🛠️ Tech Stack

  • 🐍 Python
  • ⚡ FastAPI
  • 🤖 OpenAI API
  • 🧠 Hugging Face Models
  • 🌐 Streamlit
  • 🔗 Discord Webhooks
  • 📦 JSON Processing
  • 🧰 Git & GitHub

📁 Project Structure

Sentiment-and-Urgency-detector/
│── app.py
│── requirements.txt
│── .env
│── backend/
│── ai_model/
│── utils/
│── alerts/
└── README.md

⚙️ Installation Steps

1️⃣ Clone Repository

git clone https://github.com/konathalaleela2005-collab/Sentiment-and-Urgency-detector
cd Sentiment-and-Urgency-detector

2️⃣ Create Virtual Environment

python -m venv venv

3️⃣ Activate Environment

venv\Scripts\activate

4️⃣ Install Dependencies

pip install -r requirements.txt

🔐 Environment Variables

OPENAI_API_KEY=your_api_key
DISCORD_WEBHOOK_URL=your_webhook_url
ALERT_THRESHOLD=0.75

▶️ Run Project

python app.py

OR

streamlit run app.py

📥 Sample Input

{
  "ticket": "I am extremely disappointed. If this issue is not fixed today, I will cancel my subscription."
}

📤 Sample Output

{
  "sentiment": "Negative",
  "sentiment_score": 0.94,
  "urgency_score": 0.91,
  "churn_risk_score": 0.88,
  "overall_risk": "High",
  "flagged": true
}

🚨 Discord Alert Example

🚨 HIGH PRIORITY TICKET

Risk Level: HIGH Sentiment Score: 0.94 Urgency Score: 0.91 Churn Risk Score: 0.88

Reason:

  • Negative emotional tone
  • Immediate action required
  • Customer may churn

📊 Business Impact

✔ Faster response to critical tickets ✔ Reduced customer churn ✔ Improved support efficiency ✔ Automated escalation system ✔ Better customer satisfaction

⚠️ Challenges Faced

  • Handling ambiguous customer tone
  • Prompt optimization for LLM accuracy
  • Reducing false positives
  • Real-time webhook integration

🔮 Future Enhancements

  • 🌍 Multi-language support
  • 📊 Analytics dashboard
  • 📩 Email/SMS alerts
  • 🤖 Fine-tuned domain model
  • 📈 Sentiment trend tracking

📸 Screenshots

<title>Sentiment & Urgency Detector</title> <style> * { box-sizing: border-box; margin: 0; padding: 0; } 
    body {
        font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
        background-color: #f8f9fa;
        color: #212529;
        display: flex;
        justify-content: center;
        align-items: center;
        min-height: 100vh;
        padding: 20px;
    }

    .container {
        width: 100%;
        max-width: 700px;
        background: #ffffff;
        padding: 40px;
        border-radius: 12px;
        box-shadow: 0 4px 6px rgba(0, 0, 0, 0.05);
    }

    h1 {
        font-size: 2rem;
        font-weight: 700;
        color: #1a1e21;
        margin-bottom: 24px;
    }

    .form-group {
        margin-bottom: 20px;
    }

    label {
        display: block;
        font-size: 0.85rem;
        color: #6c757d;
        margin-bottom: 8px;
        font-weight: 500;
    }

    textarea {
        width: 100%;
        height: 120px;
        padding: 16px;
        border: 1px solid #dee2e6;
        border-radius: 8px;
        background-color: #e9ecef; /* Matching the muted blue-gray tone from the image */
        font-family: inherit;
        font-size: 1rem;
        color: #495057;
        resize: vertical;
        outline: none;
        transition: border-color 0.2s ease, background-color 0.2s ease;
    }

    textarea:focus {
        background-color: #f1f3f5;
        border-color: #ced4da;
    }

    button {
        background-color: #f1f3f5;
        color: #495057;
        border: 1px solid #ced4da;
        padding: 10px 24px;
        font-size: 0.95rem;
        border-radius: 20px; /* Rounded pill style matching the image */
        cursor: pointer;
        font-weight: 500;
        transition: all 0.2s ease;
        margin-bottom: 32px;
    }

    button:hover {
        background-color: #e2e6ea;
        border-color: #adb5bd;
    }

    .result-section {
        border-top: 1px solid #e9ecef;
        padding-top: 24px;
    }

    h2 {
        font-size: 1.5rem;
        font-weight: 600;
        color: #212529;
        margin-bottom: 16px;
    }

    .result-item {
        font-size: 1rem;
        margin-bottom: 12px;
        color: #495057;
    }

    .result-value {
        font-weight: 500;
        color: #212529;
    }
</style>

Sentiment & Urgency Detector

<div class="form-group">
    <label for="complaintInput">Enter Customer Complaint</label>
    <textarea id="complaintInput" placeholder="Type customer complaint here...">I am frustrated and nobody helped me</textarea>
</div>

<button onclick="analyzeSentiment()">Analyze</button>

<div class="result-section">
    <h2>Result</h2>
    <div class="result-item">
        Sentiment: <span id="sentimentResult" class="result-value">Angry / Negative</span>
    </div>
    <div class="result-item">
        Urgency: <span id="urgencyResult" class="result-value">High</span>
    </div>
</div>
<script> function analyzeSentiment() { const text = document.getElementById('complaintInput').value.toLowerCase().trim(); const sentimentEl = document.getElementById('sentimentResult'); const urgencyEl = document.getElementById('urgencyResult'); if (text === "") { sentimentEl.textContent = "N/A"; urgencyEl.textContent = "N/A"; return; } // Mock basic client-side analysis logic for interactivity if (text.includes('frustrated') || text.includes('angry') || text.includes('nobody helped')) { sentimentEl.textContent = "Angry / Negative"; urgencyEl.textContent = "High"; } else if (text.includes('broken') || text.includes('urgent') || text.includes('help')) { sentimentEl.textContent = "Negative"; urgencyEl.textContent = "High"; } else if (text.includes('thanks') || text.includes('good') || text.includes('happy')) { sentimentEl.textContent = "Positive"; urgencyEl.textContent = "Low"; } else { sentimentEl.textContent = "Neutral"; urgencyEl.textContent = "Medium"; } } </script>

🏁 Conclusion

The Sentiment & Urgency Detector demonstrates how Artificial Intelligence can transform customer support systems by automatically prioritizing critical tickets and enabling faster decision-making through LLM-based automation.

💼 Built for Infinity Computer Solutions Placement Drive

🚀 AI-Powered Customer Support Intelligence System

About

PROJECT TITLEL: Sentiment and Urgency detector

sentiment-and-urgency-detector-dqg64ccmtf28dul7pead9m.streamlit.app/

Topics

artificial-intelligencedatamechine-learingnatural-language-processingpythonsckit-learnsentiment-analysisspacystreamlittext-classificationtraurgency-detection-classification

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages