NCA-GENL Question 27
Select 3You are tasked with building a Retrieval-Augmented Generation (RAG) system to assist customer support agents by answering queries using your company's internal knowledge base. To ensure the system is effective, which steps should you take when curating and embedding the dataset for the RAG pipeline?
- A
Clean the dataset to remove outdated, irrelevant, or redundant information.
- B
Directly embed raw documents without preprocessing to save time.
- C
Segment lengthy documents into smaller, semantically meaningful chunks before embedding.
- D
Use a domain-specific embedding model to generate vector representations.
- E
Rely exclusively on publicly available datasets to avoid managing proprietary content.
Show answer and explanation
Correct answers: A, C, D
Explanation
To create an effective RAG system, the dataset must be curated and preprocessed thoughtfully. Cleaning the data, segmenting it into manageable chunks, and using domain-specific embeddings are critical steps that improve the quality and relevance of the retrieval process. Skipping preprocessing or relying only on external datasets can lead to subpar results.
- A. Correct.
Cleaning the dataset is essential to ensure the knowledge base is accurate, reliable, and free from unnecessary noise, which improves the quality of RAG responses.
- B. Incorrect.
Directly embedding raw documents can lead to inefficient and inaccurate retrieval, as the system might struggle to identify context or relevance in unprocessed text.
- C. Correct.
Segmenting lengthy documents into smaller, meaningful chunks ensures that embeddings capture specific, relevant information, improving retrieval precision.
- D. Correct.
Using a domain-specific embedding model helps create more accurate vector representations tailored to the system's use case, enhancing performance.
- E. Incorrect.
Exclusively relying on public datasets is not suitable for proprietary or internal use cases, such as supporting company-specific queries, as it misses critical context.