CUSTOMER SERVICE CHATBOT

This project is a chatbot for an intelligent customer service chatbot. It uses a fine-tuned SentenceTransformer- based intent classification model, semantic search for a knowledge base, and logic for handling sales leads, technical support, and product feature requests.

🚀 FEATURES

• Fine-tuned intent classifier using SentenceTransformers + MLP head
• Semantic search over knowledge base article using FAISS
• Handles:
  • Technical Support queries
  • Product Feature Requests
  • Sales Lead information extraction
• Sentiment analysis via TextBlob
• Automatic escalation logging and context tracking

📁 Project Structure

project-root/
├── backend/
│   ├── main.py
│   ├── model_training/
│   │   └── intent_classifier_sentence_transfomer.pt(trained Mode)/
│   │   └── customer_intent_dataset.jsonl/
│   │   └── model_training.py/
│   ├── customer_intent_dataset.jsonl
│   ├── kb.json  
│   ├── feature_requests.json
│   ├── sales_leads.json
│   ├── negative_feedback.json
│   └── unresolved_technical_queries.json
├── frontend/
│   ├── public/
│   ├── src/
│   │   └── App.js/
│   │   └── App.css/
│   ├── package.json
│   └── ...
└── README.md

⚙️ PREREQUISITES

Please refer to the requirements.txt(backend) and requirement.txt(frontend) file in the folder having all the prerequisites for the project.

🖥 LOCAL DEPLOYMENT INSTRUCTIONS

1. Ensure the following are installed: 
   - Python 3.8+
   - A python environment manager 

2. 📦 INSTALL DEPENDENCIES 
    -pip install -r requirements.txt 
    - Initialize TextBlob data (once) 
      python -m textblob.download_corpora 



 3. 🧠 MODEL PREPARATION 
 The file intent_classifier_sentence_transformers.pt is present in the root
 directory as stated in the root directory above. This is the trained model for intent 
 classification. The model_training.py (housing the training of the model) is also in 
 model_training subfolder in the root directory.  

 
 4. 📚 Knowledge Base File
  kb.json file houses all the knowledge base with some general prompts relevant to the    
  companies 

 
 5. 🏃‍♂️ RUN THE SERVER 
    bash: 
        uvicorn main:app --reload 

    This will start the FastAPI server at: 
    http://127.0.0.1:8000

 6. 🌐 FRONTEND SETUP
     - In a different command prompt window
     - Navigating to the frontend directory 
     bash: 
         cd../frontend 
     
     - Install dependencies 
       npm install 
       npx install 
       npm install axios

     - Run the frontend dev server 
       npm start 
     - This frontend will be available at: http://localhost:3000
       
     
  7.📡 TESTING 
    - Open the browser and go to http://localhost:3000
    - While interacting with the chatbot it can be observed that 


   8. TIPS: 
    - The chatbot will log feature requests, sales leads, and unresolved tech queries 
      into JSON files. 
    - If a user expresses strong negative sentiment, the system will escalate and log 
      log their message for review in separate JSON file.