Demonstrace a využití RAG technologií pro vyhledávání v textových dokumentech

Abstract

This bachelor's thesis focuses on the design and implementation of a system for information retrieval in text documents using the Retrieval-Augmented Generation (RAG) architecture. The aim of the thesis is to combine vector-based retrieval methods with generative language models to create a system capable of providing relevant answers based on document content. The thesis describes principles of text representation using embeddings, the use of vector databases, and document processing methods, including text chunking. The solution includes the implementation of a web application that allows users to upload documents, index them, and query them using natural language. The result is a functional prototype demonstrating the advantages of the RAG approach, particularly in terms of accuracy, up-to-date information, and transparency of generated responses.