Question answering in Prompt Engineering / GenAI - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is Question Answering (QA) in AI?

Question Answering is a task where a computer program reads a question and finds the best answer from a given text or knowledge source, just like how you ask a friend and they reply with the right information.

beginner

What are the two main types of Question Answering systems?

The two main types are: 1) Closed-domain QA, which answers questions about a specific topic, and 2) Open-domain QA, which can answer questions about any topic using large knowledge sources.

intermediate

How does a machine learning model find answers in text for QA?

The model reads the question and the text, then learns to spot the exact part of the text that answers the question, similar to how you scan a book to find a sentence that answers your question.

beginner

What is the role of 'context' in Question Answering?

Context is the text or information where the answer is found. The QA system uses this context to understand the question better and find the right answer inside it.

intermediate

Name a popular dataset used to train and test Question Answering models.

SQuAD (Stanford Question Answering Dataset) is a popular dataset where models learn to answer questions by reading paragraphs from Wikipedia.

What does a Question Answering system primarily do?

A Detects objects in images

B Generates random text

C Translates languages

D Finds answers to questions from text or knowledge

QA systems focus on finding answers to questions using text or knowledge sources.

Which type of QA system answers questions about any topic?

A Open-domain QA

B Closed-domain QA

C Image QA

D Speech QA

Open-domain QA can answer questions on any topic using large knowledge bases.

In QA, what is 'context'?

A The question itself

B The text where the answer is found

C The model architecture

D The answer only

Context is the text or information where the answer is located.

Which dataset is commonly used to train QA models?

A MNIST

B ImageNet

C SQuAD

D COCO

SQuAD is a popular dataset for training and testing QA models.

How does a QA model find the answer in the text?

A By scanning the text to locate the answer span

B By translating the question

C By guessing randomly

D By summarizing the text

QA models scan the text to find the exact part that answers the question.

Explain what Question Answering is and how it works in simple terms.

Think about how you ask a friend a question and they find the answer in a book.

Describe the difference between closed-domain and open-domain Question Answering systems.

One is about specific topics, the other covers any topic.