Storytopia Hackathon Documentation
Introduction
Storytopia is an AI-powered multi-modal learning platform designed to revolutionize the way we learn by making it fun, immersive, and socially engaging. Our application utilizes the latest advancements in generative AI to transform dull educational content into vivid, engaging stories and illustrations, tailored to user preferences and learning needs.
Problem Statement
In an era of information overload, traditional learning methods such as long lectures often become tedious. Storytopia addresses this challenge by replacing complex jargon and abstract concepts with interactive, AI-generated stories and images that make learning accessible and enjoyable for everyone.
Solution Overview
Storytopia allows users to enter any topic, such as "Explain Kubernetes with an Analogy" or "Evolution of Fashion," and uses generative AI to create an immersive story illustrated in the user’s style of choice, such as comic books. Additional features include:
- Theater Mode: Automatically plays the story with AI-generated narration.
- Personalized Recommendations: Utilizes a TiDB vector database to analyze user preferences and provide tailored story suggestions, enhancing the learning experience by introducing novel topics and broadening users' horizons.
- Email Notifications: Alerts users when their story is ready.
- Social Features: Enables sharing and viewing stories, following friends, and promoting an active learning community.
- Accessibility Options: Tailors content for various disabilities, ensuring an inclusive learning environment.
Technical Architecture
- Frontend: Developed with React, providing a responsive and interactive user experience.
- Backend: Managed with FastAPI, handling API requests efficiently.
- Authentication and Hosting: Powered by Firebase, ensuring secure access and scalable hosting.
- Data Management: Utilizes Firestore for storing user and story data.
- Deployment: Firebase Hosting (frontend) and Google Cloud Run (backend) are used for reliable deployment.
- Vector Database: TiDB Serverless with Vector Search, hosted on AWS.
- Image Generation: Utilizes DALL-E 3 for creating high-quality, stylistically appropriate illustrations.
- Voice Generation: Employs Google Speech-to-Text AI to generate lifelike narrations for the stories.
- LLM Models: Incorporates Gemini-pro-1.5 for story generation and GPT-4o-mini for fast recommendation generation.
Benefits of TiDB Serverless in Storytopia
TiDB Serverless is integral to enhancing the capabilities of Storytopia:
- Semantic Search: By integrating TiDB’s vector search, we can swiftly retrieve relevant Wikipedia articles that serve as the knowledge base for generating stories.
- Retrieval-Augmented Generation (RAG): TiDB facilitates the extraction of contextual information necessary for generating high-quality, educational content.
- Scalability and Performance: Handles growing data volumes and user requests without compromising performance, crucial for live story generation and user interaction.
- Recommendation System: TiDB's vector database tracks user preferences and story interactions, enabling personalized story recommendations that adapt to user interests and encourage exploration of diverse topics.
Using TiDB for Story Generation
When a user submits a prompt, Storytopia:
- Searches for the five most relevant Wikipedia pages stored in the TiDB vector database.
- Uses the retrieved data for RAG, providing context to the Gemini-Pro-1.5 model from Vertex AI, which then generates the story.
- Stores the generated stories in a separate TiDB vector database, which maintains a profile of the user’s interests and preferences for future recommendations.
Submission Details
- TiDB Cloud Account Email: mikeguyon98@gmail.com
- Demo Video: A concise video under 4 minutes, showcasing the application’s features and the impact of TiDB Serverless, hosted on YouTube.
- Access to Application: Details provided for judges to access Storytopia with clear instructions.
- GitHub Repository: Link to the source code with complete documentation, including deployment files and architectural diagrams.
- Endpoint Swagger Documentation: the swagger docs for our api is available at Storytopia FastAPI Documentation
Conclusion
Storytopia leverages TiDB Serverless to enhance the learning experience by providing a platform that is not only educational but also engaging and inclusive. By integrating advanced AI and database technologies, Storytopia sets a new standard for educational platforms, making learning an enjoyable and community-driven experience.
Log in or sign up for Devpost to join the conversation.