Databricks Generative AI Engineer Associate Question 70
Select 3You are building a Retrieval-Augmented Generation (RAG) application to generate product support responses for a retail website. Which source documents should you prioritize to ensure the knowledge base provides accurate and high-quality answers?
- A
Product manuals and technical specifications provided by manufacturers
- B
Random customer reviews from third-party e-commerce websites
- C
Internal FAQ documents and support knowledge base
- D
Archived internal emails from the support team
- E
Competitor product descriptions from their websites
- F
Customer-submitted support tickets categorized by issue
Show answer and explanation
Correct answers: A, C, F
Explanation
To build a high-quality RAG application, it is critical to use authoritative, accurate, and relevant documents. Product manuals and internal knowledge bases provide reliable and structured information, while customer-submitted tickets can highlight real-world use cases and common issues. Other sources, like random reviews or competitor data, may introduce noise or irrelevant information.
- A. Correct.
Product manuals and technical specifications are highly reliable sources that provide detailed and accurate information about the products.
- B. Incorrect.
Random customer reviews from third-party e-commerce websites are not reliable sources as they often contain subjective opinions and may lack factual accuracy.
- C. Correct.
Internal FAQ documents and the support knowledge base are curated and authoritative sources of information that are directly relevant to customer support scenarios.
- D. Incorrect.
Archived internal emails may contain useful information but are typically unstructured and not suitable as a primary knowledge source for a RAG application.
- E. Incorrect.
Competitor product descriptions may not be relevant or accurate for the application, as they do not pertain to the retailer’s own products.
- F. Correct.
Customer-submitted support tickets categorized by issue can provide real-world examples of common problems and solutions, making them a valuable source for training the RAG application.