Similarity search and retrieval in Prompt Engineering / GenAI - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is similarity search in machine learning?

Similarity search is a method to find items that are most alike a given item, based on some measure of closeness or resemblance.

beginner

Name a common way to measure similarity between two data points.

Cosine similarity is a common measure that calculates the angle between two vectors to determine how similar they are.

beginner

Why is vector representation important in similarity search?

Vector representation converts data into numbers so computers can measure similarity using math, like distances or angles between vectors.

intermediate

What is the role of an index in similarity search and retrieval?

An index organizes data vectors so the system can quickly find the most similar items without checking every single one.

intermediate

Explain the difference between exact and approximate similarity search.

Exact search finds the perfect closest matches but can be slow for big data. Approximate search finds close matches faster but might miss the very best ones.

Which similarity measure calculates the angle between two vectors?

A Manhattan distance

B Euclidean distance

C Jaccard index

D Cosine similarity

Cosine similarity measures the cosine of the angle between two vectors, indicating how similar their directions are.

What is the main purpose of an index in similarity search?

A To train machine learning models

B To store raw data

C To speed up finding similar items

D To visualize data

An index helps quickly locate similar items without scanning all data, improving search speed.

Which of these is a drawback of exact similarity search?

A It cannot handle vectors

B It can be slow on large datasets

C It uses approximate results

D It is inaccurate

Exact search checks all data points, which can be slow when the dataset is very large.

Vector representation is important because:

A It allows mathematical comparison of data

B It stores data as text

C It removes the need for similarity measures

D It visualizes data

Vectors let computers use math to compare data points for similarity.

Which similarity measure is best for comparing sets of items?

A Jaccard index

B Cosine similarity

C Euclidean distance

D Pearson correlation

Jaccard index measures similarity between sets by comparing shared and total items.

Describe how similarity search works and why it is useful in real life.

Think about how online stores suggest products you might like.

Explain the difference between exact and approximate similarity search and when you might use each.

Consider searching a huge photo collection quickly.