Checklist NOAI Singapore 2025 Final · Task 3
Retrieval Augmented Generation
Build a minimal TF-IDF retrieval step over five sentences and compose a response to a query.
The task
Question 3 (40 marks) frames a classroom story in which a smart assistant must answer 'How does machine learning work?' from five one-sentence documents. The contestant vectorises the documents (for example with TF-IDF), ranks them by cosine similarity, prints the retrieved documents in top_documents and a generated response, and explains the thought process. The expected response combines the second and fourth documents.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
- Five documents and a query (fixed in the notebook).
- You submit
top_documents, a response string and a written explanation.- Rules
- Python 3.9 standard library and NumPy only.
- Format
- NOAI 2025 competition day, 8 March 2025, NTU College of Computing and Data Science; 2.5 hours; Section 1 (MCQs) and Section 2 (three programming questions, 80 marks).