Databricks Generative AI Engineer Associate Question 223
Select 3You are tasked with building a document search application using Databricks. You need to create and query a Vector Search index to retrieve the most relevant documents based on user queries. Which of the following steps are necessary to implement this solution correctly?
- A
Generate vector embeddings for the documents using a pre-trained model.
- B
Create a Vector Search index with the generated embeddings.
- C
Store the original documents in the index instead of their embeddings.
- D
Use cosine similarity or another similarity metric to retrieve nearest neighbors during query execution.
- E
Directly query the raw text documents without generating embeddings.
Show answer and explanation
Correct answers: A, B, D
Explanation
To implement a Vector Search solution in Databricks, you need to generate embeddings for the documents, create a Vector Search index to store these embeddings, and use a similarity metric such as cosine similarity to retrieve relevant results. Storing original documents or querying raw text directly is not part of the process, as embeddings are the core representation for semantic search.
- A. Correct.
Correct: Generating vector embeddings for documents is essential to represent their semantic meaning in a format that can be indexed and queried.
- B. Correct.
Correct: Creating a Vector Search index is necessary to store and organize the embeddings for efficient retrieval.
- C. Incorrect.
Incorrect: The original documents are not stored in the Vector Search index. Instead, embeddings (numerical representations of the documents) are stored.
- D. Correct.
Correct: Using a similarity metric like cosine similarity is required to find the most relevant embeddings during the query process.
- E. Incorrect.
Incorrect: Raw text documents cannot be directly queried in a Vector Search system. Embeddings are needed to enable semantic search capabilities.