Question 1

How do vector databases enhance recommendation systems?

❌ Simplify complex queries
✅ Enable efficient similarity searches for complex data types
❌ Provide more secure storage
❌ Support larger data volumes

Explanation:
Vector DBs perform fast similarity search on embeddings, ideal for recommendations.

---

Question 2

Key difference between vector DBs and traditional DBs:

❌ Traditional DBs optimized for numerical data
✅ Vector databases are designed to handle embeddings
❌ Vector DBs use relational tables
❌ Traditional DBs best for similarity search

Explanation:
Embeddings are high-dimensional vectors—vector DBs are built to store & search them efficiently.

---

Question 3

Which differentiates in-memory vector DBs?

✅ Store vectors directly in memory → faster queries
❌ Store vectors on disk
❌ Slower than others
❌ Do not support vector search

Explanation:
In-memory storage gives very fast read/query times.

---

Question 4

Which metric measures direction, not magnitude?

❌ Inner product
✅ Cosine distance
❌ Euclidean
❌ Dot product

Explanation:
Cosine distance evaluates the angle between vectors, ignoring magnitude.

---

Question 5

What does the direction of an embedding represent?

❌ Distance between unrelated items
❌ Frequency in dataset
✅ Topic or semantic meaning
❌ Strength of the item

Explanation:
Direction encodes semantic meaning, e.g., topics or conceptual relationships.

---

Question 6

Indexing vs Querying in Chroma DB:

❌ Indexing = input, querying = output
❌ Serve same purpose
✅ Indexing organizes data; querying retrieves based on semantic similarity
❌ Encrypt/decrypt data

Explanation:
Indexing builds efficient structures; querying finds nearest neighbor vectors.

---

Question 7

Role of embeddings in similarity search:

❌ Encrypt data
❌ Store metadata
❌ Increase database speed
✅ Convert data into a vector format that enables similarity comparison

Explanation:
Embeddings transform text/images/etc. into vectors so similarity algorithms can compare them.

---

🧾 Summary Table