Question Answering Systems (Extractive and Generative) Extractive Question Answering An extractive question answering system relies on existing knowledg...
Question Answering Systems (Extractive and Generative)
Extractive Question Answering
An extractive question answering system relies on existing knowledge resources to generate the best answer to a question. The system scans through a set of pre-defined documents and extracts relevant information, which is then used to generate the answer.
Examples:
Question: What is the main character in the book "The Iliad"?
Answer: Achilles, a Trojan prince.
Sources: The Iliad, various historical texts.
Generative Question Answering
A generative question answering system is capable of generating new answers to questions that it has not encountered before. This type of system uses natural language processing (NLP) techniques to understand the meaning of the question and generate a relevant answer.
Examples:
Question: Write a poem about the beauty of nature.
Answer: The vibrant hues of nature paint a canvas of beauty, inviting us to marvel at the wonders that surround us.
Source: A generative language model trained on vast amounts of literature.
Key Differences between Extractive and Generative QA
| Feature | Extractive QA | Generative QA |
|---|---|---|
| Knowledge source | Pre-defined documents | None |
| Answer generation | Extracted information | Generated |
| Context dependence | Limited | Contextual |
| Example | Extracting the name of a character | Generating a poem about a person |