Wiki LLM RAG
Explore the novel Wiki LLM paradigm for Retrieval Augmented Generation (RAG).
Requirements
- M.Sc. in Machine Learning, Data Science, Computer Science, Mathematics, Telecommunications, or similar
- Knowledge of Python
- Software development skills
- Knowledge of natural language understanding
- Basic knowledge LLM, RAG, knowledge graphs
Description
This thesis/internship investigates the LLM Wiki paradigm as an alternative and complement to traditional Retrieval Augmented Generation. Instead of retrieving raw document chunks at query time, the project explores how an LLM can incrementally build and maintain a persistent, interlinked wiki from curated sources. New documents are processed into structured markdown pages, entity and concept summaries are updated over time, cross-references are maintained, and contradictions or stale claims are flagged. The goal is to design and prototype a system where knowledge accumulates as a durable artifact, enabling more efficient, transparent, and context-aware question answering over evolving collections of documents.