AI-102 Question 487
Select 2You are developing a chatbot powered by Azure OpenAI and need to ensure it can answer questions based on your organization's proprietary documents stored in Azure Blob Storage. Which steps should you take to integrate your data with the Azure OpenAI model effectively?
- A
Store your documents in Azure Blob Storage and enable Cognitive Search to index the content.
- B
Create a custom knowledge base using Azure Cognitive Services QnA Maker and link it to the OpenAI model.
- C
Configure Azure Cognitive Search to use a vector-based search index to improve the relevance of results.
- D
Directly upload the documents to the Azure OpenAI service for training the model.
- E
Use Azure Logic Apps to automate the ingestion of data into the OpenAI model.
Show answer and explanation
Correct answers: A, C
Explanation
To use your proprietary data with an Azure OpenAI model, you must store the data in Azure Blob Storage and leverage Azure Cognitive Search to index and retrieve this data. This allows the model to perform retrieval-augmented generation (RAG) by querying the indexed data during inference. Using vector-based search further enhances the relevance of retrieved results, making it easier for the model to provide accurate and context-aware answers.
- A. Correct.
Correct: Storing documents in Azure Blob Storage and enabling Cognitive Search is a key step for integrating proprietary data. Cognitive Search creates an index from the documents, allowing the OpenAI model to retrieve relevant information during inference.
- B. Incorrect.
Incorrect: QnA Maker is a separate service and does not directly integrate with Azure OpenAI models. While it can be used to create a knowledge base, this is not the correct approach for integrating proprietary documents with the OpenAI model.
- C. Correct.
Correct: Using a vector-based search index in Azure Cognitive Search improves the relevance of the results returned when querying the proprietary data. This is crucial for enhancing the model's ability to retrieve accurate information.
- D. Incorrect.
Incorrect: Azure OpenAI does not allow direct document uploads for model training. Instead, it relies on external data sources like Azure Blob Storage and Cognitive Search for integrating custom data.
- E. Incorrect.
Incorrect: Azure Logic Apps can automate workflows but is not a method for integrating data directly with the OpenAI model.