Module 03 Knowledge Check
5 questions. Pass mark 4 out of 5. Answer every question before checking the answer key below, then retry after reading the feedback.
1. What does an embedding represent?
- A. A compressed copy of the text
- B. A numeric position capturing similarity of meaning
- C. The document's permissions
- D. A hash for deduplication
2. Semantic search returns a passage that is topically similar but does not answer the question. This shows…
- A. The embedding model is broken
- B. Similarity is not the same as answerhood
- C. The index is corrupt
- D. The chunk is too long
3. When does keyword (lexical) search still beat pure vector search?
- A. Never
- B. For exact identifiers - product codes, error numbers, names
- C. For long questions
- D. For images
4. Increasing top-k retrieved chunks tends to…
- A. Always improve answers
- B. Raise recall but add distracting context that can worsen the answer
- C. Reduce cost
- D. Fix chunking problems
5. Re-ranking after retrieval is used to…
- A. Compress the index
- B. Reorder candidates with a more accurate (costlier) relevance model before answering
- C. Encrypt results
- D. Translate documents
---
Answer Key and Explanations
Check these only after attempting every question.
1. B - A numeric position capturing similarity of meaning
Embeddings place similar meanings near each other. They are not reversible copies and carry no access rules.
2. B - Similarity is not the same as answerhood
Vector search optimises closeness of meaning, not whether the passage actually contains the answer.
3. B - For exact identifiers - product codes, error numbers, names
Rare exact tokens are precisely what embeddings blur. Hybrid retrieval exists for this reason.
4. B - Raise recall but add distracting context that can worsen the answer
More context is not automatically better; irrelevant passages measurably degrade grounded answering.
5. B - Reorder candidates with a more accurate (costlier) relevance model before answering
Cheap recall first, expensive precision second - a standard two-stage retrieval pattern.
Practical Check
Apply this module to your own work: complete the module activity for *Embeddings, Indexes, and Retrieval in Plain English*, then write one sentence naming what your result shows and one naming what it does not.
Strong Answer Pattern
A strong answer names the task, the evidence used, the check performed, and the remaining limitation. It avoids "proved", "guaranteed", or "always" unless the evidence genuinely supports it.