Fact-Checking for LLM-generated Scientific References

Large Language Models (LLMs) can generate scientific texts and references. Sometimes, they “hallucinate” and generate references that do not actually exist. The goal of this thesis project is to design and implement a fact-checker for LLM-generated scientific references that can extract and parse references from plain text and PDF and check if they actually exist by utilising publicly available resources like dblp or DOIs.