Implementing rag using graphs to improve information retrieval

Identifier:  TFG:9457

Handle: https://hdl.handle.net/20.500.11797/TFG9457

Authors:  Aragoneses Romero, Anna

Abstract:
It presents the implementation of a Retrieval-Augmented Generation (RAG) system based on knowledge graphs to improve the retrieval and generation of information from unstructured documents, such as technical PDFs. The process includes the extraction of entities, attributes and relationships with Amazon Comprehend and spaCy, hierarchical structuring in Neo4j, and automatic quality verification using an LLM agent. Finally, the RAG module is integrated to provide more precise answers. The results show that the use of graphs increases the interpretability and relevance of the answers.