Hexaware CODE&RISE PROGRAM
GENMIND
Perfecting the Science of Questioning
Team Details
Team Name: GENMIND
Application Name: Automated Question Builder Application
S.No TEAM MEMBER’S NAME EMAIL
© Hexaware Technologies. Confidential briefing. 2
Impact/Potential Value of the Application
Our Automated Question Builder application leverages advanced AI technology to streamline and enhance
the creation of educational content. By automating the generation of questions and answers from PDF documents,
this tool significantly reduces the time and effort required for educators and trainers to develop comprehensive
assessment materials. The application is scalable across various domains, making it valuable for both corporate
training environments and educational institutions.
• Efficiency in Content Creation:
• Automates the generation of educational content, significantly reducing manual effort.
• Enables educators and trainers to focus on teaching and mentoring rather than content creation.
• Scalability Across Domains:
• Useful in corporate training, academic institutions, and professional certification programs.
• Can be customized to different subjects, industries, and levels of difficulty.
• Cost-Effective Solution:
• Reduces the need for expensive content creation services.
• Provides a cost-effective tool for creating tailored educational materials.
• Enhanced Learning Outcomes:
• Generates content that aligns with the curriculum, ensuring relevance and improving learning
outcomes.
• Offers different difficulty levels, catering to diverse learning needs and abilities.
The Solution Proposed by your Team
We propose an AI-powered application that automates the extraction of content
from PDF documents and generates relevant questions and answers. The tool uses
natural language processing (NLP) and machine learning (ML) models to understand the
context and generate high-quality educational content that aligns with specified
difficulty levels.
Solution Highlights Key Features / Approach
Automated Content Generation: The application PDF Text Extraction: Efficiently extracts text from PDF
automates the creation of questions and answers from PDF documents.
documents, saving significant time for educators.
Customizable Difficulty Levels: Users can select easy, Contextual Question Generation: Uses context from the
medium, or hard difficulty levels for the questions generated. extracted text to generate relevant questions and answers.
AI-Driven Insights: The application leverages Google Difficulty Classification: Classifies questions into easy,
Generative AI to ensure the relevance and accuracy of the medium, and hard based on predefined keywords and
content generated. complexity.
User-Friendly Interface: Built with Streamlit, providing an
intuitive interface for users to interact with the application.
© Hexaware Technologies. Confidential briefing. 4
Technologies Used
Frontend:
HTML5: Structuring the web content.
CSS3: Styling the user interface, ensuring a modern and professional look.
JavaScript: Adding interactivity and dynamic features to the application.
Streamlit: Providing an easy-to-use framework for building the frontend in Python.
Backend:
Python: The core language used for backend development.
PyPDF2: For extracting text from PDF documents.
Google Generative AI: For embedding generation, similarity search, and
conversational AI.
FAISS: For efficient storage and retrieval of vectorized text data.
APIs and External Services:
Google Generative AI API: For embedding and conversational AI functionalities.
Environment Variables: Managed using Python's dotenv for secure API key handling.
© Hexaware Technologies. Confidential briefing. 5
Gen AI Tool Utilization
Our application integrates Google Generative AI for both embeddings and conversational AI
functionalities. The embeddings model is used to represent the text data in a vectorized form,
enabling efficient similarity search and context-aware question generation. Additionally, Google’s
conversational AI model is employed to generate detailed and contextually accurate answers to the
questions generated.
1. Google Generative AI for Embeddings:
1. Converts extracted text into vector embeddings, enabling similarity searches.
2. Ensures that the context of the text is preserved, leading to more accurate question generation.
2. Google Generative AI for Conversational AI:
1. Generates detailed answers to questions based on the provided context.
2. Enhances the user experience by providing accurate and contextually relevant responses.
3. Custom Prompt Engineering:
1. Uses a carefully crafted prompt template to guide the AI in generating the most relevant and detailed answers.
2. Ensures that the AI only provides answers when the context is available, avoiding incorrect or misleading
information.
© Hexaware Technologies. Confidential briefing. 6
System Architecture, Functionalities and Design Diagram
© Hexaware Technologies. Confidential briefing. 7
System Architecture, Functionalities and Design Diagram
Data Flow:
• PDF Upload: Users upload PDF documents via the Streamlit interface.
• Text Extraction: The system extracts text from the uploaded PDFs using PyPDF2.
• Text Chunking: The extracted text is split into chunks for processing using a Recursive Character Text Splitter.
• Vectorization: Text chunks are converted into vector embeddings using Google Generative AI.
• Storage in FAISS: The vectorized data is stored in FAISS for efficient similarity search.
• Question Generation: The system generates questions from the text chunks.
• Answer Generation: Detailed answers are provided using Google Generative AI’s conversational model.
Components and Interactions:
• Frontend (Streamlit, HTML5, CSS3, JavaScript): User uploads PDFs and interacts with the generated questions
and answers.
• Backend (Python, Google Generative AI, FAISS): Handles text processing, vectorization, storage, and AI-
based question-answering.
API and External Services:
• Google Generative AI API: Used for both embedding generation and conversational AI.
• FAISS: Used for storing and searching vectorized data.
• Environment Management: API keys and sensitive information securely managed using .env files.
© Hexaware Technologies. Confidential briefing. 8
How it works
© Hexaware Technologies. Confidential briefing. 9
How It Works
Technical Architecture
Frontend:
• Streamlit Framework: Simplifies the deployment of the application and
integration of Python-based processing.
• HTML5 and CSS3: Ensures a responsive and visually appealing user interface.
• JavaScript: Adds dynamic functionality, enhancing user interaction.
Backend:
• Python Scripting: Core of the backend, handling data processing and AI
interaction.
• API Integration: Utilizes Google Generative AI APIs for embedding
and conversational models.
• Data Storage: FAISS is used to store and retrieve vectorized text data efficiently.
Security and Performance:
• Environment Variables: API keys are securely managed using Python’s dotenv.
• Optimized Processing: Efficient algorithms ensure quick text
processing and question generation.
© Hexaware Technologies. Confidential briefing. 10
Innovation and Creativity
• AI-Driven Content Generation:
• Uses cutting-edge AI to automatically generate educational content from existing materials.
• Provides an innovative solution to the time-consuming process of manual content creation.
• Personalization and Customization:
• Offers customizable difficulty levels, making the tool adaptable for different
learning
environments.
• Allows users to focus on specific sections or topics, providing tailored content generation.
• Streamlined User Experience:
• Combines the power of AI with a user-friendly interface, making advanced
technology
accessible to non-technical users.
Scalability, Performance and Security
Scalability:
• Designed to handle large datasets and multiple concurrent users.
• Can be scaled to support enterprise-level operations in corporate training and
educational institutions.
Performance:
• Optimized for fast text processing, ensuring minimal latency even with large documents.
• Efficient storage and retrieval using FAISS, enabling quick access to processed data.
Security:
• Secure handling of API keys and sensitive information using environment variables.
• Adheres to industry standards for data privacy and security, ensuring that user
data is protected.
© Hexaware Technologies. Confidential briefing. 12
Best practices and industry standards followed
Coding Standards:
• Adheres to Python PEP 8 coding standards, ensuring clean and maintainable code.
• Follows best practices for HTML5, CSS3, and JavaScript to ensure a responsive and accessible
frontend.
Data Privacy and Security:
• API keys are managed using .env files, protecting sensitive information.
• Implements secure handling of user-uploaded data, ensuring compliance with data
protection regulations.
Scalability and Performance:
• Designed with scalability in mind, ensuring the application can handle growing data and
user demands.
• Optimized algorithms for efficient processing, providing fast and reliable performance.
© Hexaware Technologies. Confidential briefing. 13
User Experience
Intuitive Interface:
• Streamlined Navigation: Easy-to-use interface with clear workflows.
• Minimal Learning Curve: Accessible for users with varying technical skills.
Responsive Design:
• Cross-Device Compatibility: Optimized for desktops, tablets, and smartphones.
• Fast Load Times: Quick processing and response, even with large documents.
Customization:
• Personalized Questions: Users select difficulty levels (easy, medium, hard).
• Focused Content: Generate questions from specific sections of text.
Real-Time Interaction:
• Instant Results: Immediate question generation and answers.
• Interactive Experience: Engage with questions and receive detailed explanations.
Accessibility Features:
• Keyboard Navigation & High Contrast Mode: Ensures accessibility for all
users.
© Hexaware Technologies. Confidential briefing. 14
Console Output Details
BASIC STRAMLIT OUT PUT OF GENAI (GOOGLE GEMINI PRO) FOR AUTOMATIC QUESTI0N GENERATION.
© Hexaware Technologies. Confidential briefing. 15
Console Output Details
BASIC STRAMLIT OUT PUT OF GENAI (GOOGLE GEMINI PRO) FOR AUTOMATIC QUESTI0N GENERATION.
© Hexaware Technologies. Confidential briefing. 16
Thank You
