Skip to content

Latest commit

 

History

History
121 lines (93 loc) · 5.41 KB

azure_chatbot_customer_interaction.md

File metadata and controls

121 lines (93 loc) · 5.41 KB

Azure Chatbot for Automated Customer Interaction

Introduction

This project, carried out during my internship as a Project Intern on Microsoft Azure at Verzeo in July 2019, focuses on developing and deploying a chatbot using Microsoft Azure Bot Service. The chatbot, powered by QnA Maker, serves as an intelligent conversational agent capable of answering predefined questions and integrating with web pages and platforms like Facebook Messenger. The solution demonstrates the creation of a scalable, interactive, and AI-driven bot using Azure's AI + Machine Learning services.

The presentation for this project is publicly available at Deploying a Chatbot Service Using Microsoft Azure.

Objectives

  1. Design and Deploy a Chatbot:
    • Create a web-based chatbot using Azure's Bot Service and QnA Maker.
  2. Knowledge Base Creation:
    • Automate responses by linking the bot to a curated knowledge base (FAQs).
  3. Integration with External Platforms:
    • Enable chatbot deployment on websites and messaging platforms.
  4. Real-Time Testing and Optimization:
    • Evaluate the chatbot's functionality through web-based simulations.

System Architecture

1. Components

  • QnA Maker: Service to create, train, and deploy a knowledge base for FAQs.
  • Azure Bot Service: Framework to host and deploy the chatbot.
  • Web App Bot: Azure-hosted web application to manage and test the bot.
  • Direct Line API: Facilitates communication between the bot and client applications.
  • LUIS Integration (Optional): Supports natural language processing for advanced conversational capabilities.

2. Workflow

  1. Knowledge Base Creation:
    • Curate FAQs from a URL or upload files (e.g., PDFs, Excel sheets).
    • Use QnA Maker to structure the question-answer pairs.
  2. Bot Deployment:
    • Host the bot on Azure as a Web App Bot under the AI + Machine Learning domain.
  3. Integration and Testing:
    • Integrate the chatbot into platforms like web pages or Facebook Messenger.
    • Test the chatbot in the Azure Web Chat interface.

Implementation Steps

1. Knowledge Base Creation

  1. Visit QnA Maker Portal:
    Navigate to QnA Maker and select "Create a Knowledge Base."
  2. Create QnA Service:
    Link the service to an Azure resource by selecting "Create a QnA Service."
    Fill in required details like resource group, pricing tier, and region.
  3. Upload Data Source:
    • Use a URL (e.g., Microsoft Azure FAQs) or upload local files (PDF, Excel) as a data source.
  4. Save and Train:
    • Curate or add custom QnA pairs to the knowledge base.
    • Save changes and train the model for accuracy.
  5. Pre-Test Knowledge Base:
    • Test the curated QnA pairs to verify functionality.

2. Deploying the Bot

  1. Visit Azure Marketplace:
    Navigate to the AI + Machine Learning domain in the Azure portal.
  2. Create a Web App Bot:
    • Select "Web App Bot" and provide the required inputs:
      • Bot Name: e.g., MinorChatBot.
      • Region: Based on user location.
      • App Service Plan: Choose a scalable pricing tier.
    • Click "Create" to deploy the bot.
  3. Test in Web Chat:
    • After deployment, test the bot using the "Test in Web Chat" feature to validate its functionality with the connected QnA knowledge base.

3. Integration and Testing

  1. Integrate Platforms:
    • Deploy the chatbot on web pages, Facebook Messenger, or other platforms using Direct Line API.
  2. Testing:
    • Use the Azure Web Chat interface and integrated platforms to validate the bot's responses.

Key Features

  1. Dynamic Knowledge Base:
    • Supports custom and curated QnA pairs for varied use cases.
  2. Platform Integration:
    • Compatible with web and social media platforms via APIs.
  3. Scalability:
    • Hosted on Azure with automatic scaling for handling high traffic.
  4. Advanced Features (Optional):
    • Integration with LUIS for context-aware and natural language processing capabilities.

Learning Outcomes

  1. Cloud-Based Chatbot Development:
    • Gained expertise in using Azure services for chatbot creation.
  2. QnA Maker Integration:
    • Learned to structure and train a knowledge base for automated responses.
  3. Web App Deployment:
    • Acquired skills in deploying and managing Azure-hosted applications.
  4. Direct Line API Implementation:
    • Developed an understanding of API integration for external platform connectivity.
  5. Real-Time Testing:
    • Enhanced skills in testing and optimizing chatbot performance for diverse use cases.

Conclusion

This project showcases the design and deployment of a chatbot using Microsoft Azure Bot Service and QnA Maker. The chatbot automates responses, improving user engagement and operational efficiency. By integrating with web platforms, the system demonstrates versatility and scalability, making it suitable for a wide range of applications.

Author: Yashas Javali
Duration: July 2019
Specialisation: Cloud Computing and AI Integration
Statement: This project reflects the practical application of Azure Bot Service and QnA Maker, showcasing their capabilities in creating intelligent conversational agents for real-world use cases.