Extractive QA concept in NLP - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is Extractive Question Answering (QA)?

Extractive QA is a type of question answering where the answer is found by selecting a span of text directly from a given passage or document.

beginner

How does Extractive QA differ from Generative QA?

Extractive QA picks answers directly from the text, while Generative QA creates answers in its own words, possibly not found exactly in the text.

beginner

What is the typical input for an Extractive QA model?

The input usually includes a question and a context passage where the answer is expected to be found.

intermediate

Which machine learning models are commonly used for Extractive QA?

Models like BERT and its variants are commonly used because they understand context well and can find answer spans in text.

intermediate

What are start and end tokens in Extractive QA?

They are positions in the text that mark where the answer begins and ends, helping the model select the exact answer span.

In Extractive QA, where does the answer come from?

A From user input

B Generated from scratch by the model

C Directly from the given text passage

D From a database lookup

Extractive QA selects the answer span directly from the provided text passage.

What does an Extractive QA model predict to find the answer?

A A generated sentence unrelated to the passage

B A summary of the passage

C A yes or no answer

D Start and end positions of the answer in the text

The model predicts the start and end tokens to extract the answer span.

Which of these is a popular model architecture for Extractive QA?

A K-Means

B BERT

C GAN

D RNN without attention

BERT is widely used for Extractive QA due to its strong contextual understanding.

What is the main difference between Extractive and Generative QA?

A Extractive selects text from passage; Generative creates new text

B Extractive creates new text; Generative selects from passage

C Both generate new text

D Both select text from passage

Extractive QA picks answers from the passage, Generative QA creates answers.

What inputs does an Extractive QA system need?

A A question and a context passage

B Only a question

C Only a passage

D A question and an answer

Extractive QA requires both a question and a passage to find the answer.

Explain how an Extractive QA model finds an answer in a text passage.

Think about how the model points to the exact words in the text.

Describe the difference between Extractive and Generative QA in simple terms.

Consider whether the answer is copied or made up.