Asking for advice on my approach for following Wikipedia project: Create a Classification system and then build a Knowledge Graph to help Wikipedia search, maybe use for search in LLM RAG tools

Hi




I work for Wikimedia/Wikipedia. I'm looking for suggestions and recommendations (and what to avoid) in ML for a NLP project.




Task:




Make Wikipedia search easier and add better structured data for Wikipedia




Phase 1 - Build a simple multi class classification system




Initial idea is to build a hybrid ML + heuristic rules to define a set of 30—80 top-level topic classes and tag each article with a set of class labels (with confidence scores).










Phase 2 - Add …

advice build classification graph knowledge knowledge graph languagetechnology llm llm rag nlp project rag recommendations search suggestions tools wikipedia work