Blazor PDF Chat

A Blazor web app demonstrating interaction with PDF documents using AI through a Retrieval-Augmented Generation (RAG) approach.

Overview

This project showcases:

• Document Processing: Load, extract, and chunk PDF documents.
• Vectorization: Convert text chunks into vectors for efficient similarity search.
• Vector Storage: Store these vectors in a vector database.
• Query Handling: When a user asks a question:
  1. The query is vectorized.
  2. A search for the nearest neighbor matches in the vector database is performed using cosine similarity.
  3. The query prompt is augmented with relevant document snippets to provide accurate responses.

Key Features

• Blazor Implementation: Frontend built with Blazor for a rich, interactive user experience.
• Semantic Kernel Vector Store: Utilizes a Semantic Kernel in-memory vector store implementation, but you can drop in a different vector store connector.
• Flexible AI Configuration: Leverages PredictionGuard.NET NuGet package, but you can drop in another Microsoft.Extensions.AI abstraction.