Answer: Dense passage retriever followed by a generative model (e.g., GPT-based)

## Explanation **B. Dense passage retriever followed by a generative model (e.g., GPT-based)** This combination is ideal for building a system that retrieves relevant information from a document repository and generates natural language answers. The dense passage retriever uses dense embeddings to search and retrieve the most relevant documents, and the generative model produces fluent, context-aware natural language responses based on those documents. ### Why this works best: - **Dense passage retriever**: Uses semantic similarity through embeddings to find documents that are contextually relevant to the user's question, not just keyword matches - **Generative model (e.g., GPT-based)**: Can synthesize information from multiple retrieved documents and generate coherent, natural language answers that directly address the user's query ### Why other options are less suitable: - **A**: NER followed by summarization doesn't provide effective document retrieval and lacks the generative capabilities for question-answering - **C**: Text classification followed by generation doesn't handle the document retrieval aspect effectively - **D**: Keyword-based search is less effective than semantic search for understanding natural language questions

Author: LeetQuiz .