Large Language Models (LLMs) are currently in high demand as they offer opportunities for faster content retrieval, generation, and summarization. Nonetheless, several challenges, such as bias, inaccuracies, and hallucinations, need to be addressed. Additionally, the lack of domain-specific knowledge and the inability to validate and attribute sources hinder widespread adoption.

Graph databases are a great tool for representing and manipulating information in the form of knowledge graphs. Combining Knowledge graphs with LLMs, enhances accuracy, explainability, and context in LLM-generated answers. Therefore, Knowledge graphs are an ideal complement to LLMs, ensuring accuracy and context in their outputs.

This project will use a graph database Neo4J to build a knowledge graph and combine it with LLM to demonstrate and evaluate Question/Answering tasks. In particular, the project will focus the knowledge graph on geographic places from data in Wikidata/Wikipedia.

You will benefit for a good library of resources and learning courses on the subject provided by Neo4j. https://neo4j.com/generativeai/ Keywords: Graph databases, Large-scale data management, Python and Machine Learning