AI-102 Question 365
Select 2You are designing an Azure Cognitive Search solution to process unstructured text documents stored in Azure Blob Storage. The documents are in multiple languages, and you need to extract language-specific key phrases and named entities as part of the indexing process. How should you define the skillset to meet these requirements?
- A
Include a Text Analytics skill to detect the language of the documents.
- B
Add a custom skill to pre-process the documents before extracting entities.
- C
Use a Text Analytics skill to extract key phrases and named entities for each document.
- D
Add an Image Analysis skill to extract text from any embedded images.
- E
Configure the skillset to include only language detection, as key phrase extraction is handled by the indexer automatically.
Show answer and explanation
Correct answers: A, C
Explanation
To process unstructured multilingual documents in Azure Cognitive Search, a skillset must include a Text Analytics skill to detect the language and extract relevant information like key phrases and named entities. These are required steps to ensure the documents are properly indexed with language-aware extraction of entities and phrases.
- A. Correct.
Correct. Language detection is necessary to process multilingual documents and ensure the correct language model is applied for key phrase and named entity extraction.
- B. Incorrect.
Incorrect. A custom skill is not required in this scenario since standard Cognitive Search skills like language detection and key phrase extraction are sufficient.
- C. Correct.
Correct. Extracting key phrases and named entities requires the use of the Text Analytics skill, which is a built-in skill for Cognitive Search.
- D. Incorrect.
Incorrect. Image Analysis is not relevant in this scenario since the requirement is to process unstructured text documents, not images.
- E. Incorrect.
Incorrect. While language detection is a necessary step, key phrase extraction and named entity recognition require explicit skill configuration and are not handled automatically by the indexer.