Weekend Challenge · Week 2
Break BM25
The fastest way to understand a system is to find its edges. Your mission: make your own search engine fail, on purpose, five different ways.
Your mission
- Find 5 queries where your BM25 baseline retrieves the wrong documents from your corpus
- Diagnose each failure: synonymy, paraphrase, misspelling, question-vs-statement phrasing, or something else
- Predict, in writing, which failures Week 3's embeddings will fix
Deliverable
bm25_failures.md with the 5 queries, diagnoses, and predictions, committed.
Stretch goal
Try to fix one failure without embeddings — synonym expansion, stemming, or query rewriting. How far can classical tricks go?