AI-102 Implement knowledge mining and information extraction solutions Practice Test 5 — 15 Questions

Question 1

Which TWO capabilities are available in Azure AI Search to improve search relevance? (Choose two.)

Accepted Answer

Scoring profiles. Scoring profiles (C) are correct because they let you boost the relevance of results by weighting specific fields, applying functions (such as magnitude, freshness, distance, or tags), and combining them with boost values in the scoring profile. Semantic ranking (D) is correct because it uses Microsoft's language understanding models to re-rank the top results from the initial BM25/Lucene ranking, promoting results that are semantically closer to the query's intent. Filters (A) are used to narrow results by exact criteria (for example, $filter on a field) and do not affect relevance scoring. Indexers (B) are used to automate data ingestion and enrichment from sources into an index, not to improve ranking. Synonym maps (E) expand queries with equivalent terms to improve recall, but they are a query-expansion feature rather than a relevance-ranking capability.

Answer

Filters

Answer

Indexers

Answer

Synonym maps

Question 2

Refer to the exhibit. You have this skillset definition for an Azure AI Search enrichment pipeline. You notice that the entity recognition skill is not executing on any document. What is the most likely cause?

Accepted Answer

The input source path in the entity recognition skill should be relative to the context, not absolute. In an Azure AI Search skillset, each skill's input source path is resolved relative to the enrichment context (the /document or /document/pages/* node), not as an absolute path. If the entity recognition skill's input uses an absolute path like /document/content instead of a context-relative path such as /document/pages/*/content, the skill receives no input and silently produces no output. Correcting the source to be relative to the defined context restores execution.

Answer

The entity recognition skill requires a language code that is missing

Answer

The split skill is not producing pages because the content is too short

Answer

The entity recognition skill is not registered in the skillset

Question 3

You are designing an Azure Cognitive Search solution for a legal document repository. The solution must support search across multiple languages, including English and Spanish. You need to ensure that searches return documents regardless of the language of the query. Which approach should you use?

Accepted Answer

Create separate fields for each language and set the analyzer per field.. The recommended approach in Azure AI Search for multilingual search is to create separate fields for each language and configure each field with the appropriate language analyzer (e.g., Microsoft English analyzer for English text, Microsoft Spanish analyzer for Spanish text). When querying, the client application can determine the query language and target the corresponding field, or use a `searchFields` parameter to search across multiple language-specific fields. This ensures that language-specific linguistic processing (stemming, stop words, etc.) is applied correctly. Option C is incorrect because Azure AI Search does not allow assigning multiple analyzers to a single field; a field can have only one analyzer. Options A and B are also incorrect: a custom splitter does not provide language-specific analysis, and assuming translation is unreliable.

Answer

Create a custom analyzer that splits text by language.

Answer

Use a single language analyzer set to English, and assume Spanish documents are translated.

Answer

Use the Microsoft analyzer for English and Spanish in a multi-language field.

Question 4

You are implementing a knowledge mining solution using Azure Cognitive Search with built-in AI enrichment. The pipeline must extract named entities and key phrases from documents. The enrichment pipeline should be triggered only for documents that are larger than 1 MB. Which approach should you use?

Accepted Answer

Implement a custom skill that checks the document size and only calls the built-in skills if size > 1 MB.. Azure Cognitive Search does not natively support document-size-based conditional skill execution. By implementing a custom skill that checks the document size and only invokes the built-in entity recognition and key phrase extraction skills for documents larger than 1 MB, you achieve the required conditional enrichment logic. This approach leverages the custom skill's ability to inspect document metadata and control downstream skill execution.

Answer

Configure the skillset to run entity recognition and key phrase extraction on all documents.

Answer

Use indexer parameters to specify a document size filter.

Answer

Create two indexers: one for large documents with skills, one for small documents without.

Question 5

You are designing an Azure Cognitive Search solution that indexes customer support tickets. The index must include a field for 'sentiment' that is populated from an AI enrichment pipeline. Which TWO actions are required to achieve this?

Accepted Answer

Add a built-in Sentiment skill to the skillset.. Option A is correct because the built-in Sentiment skill (Microsoft.Skills.Text.SentimentSkill, or the V3 equivalent Microsoft.Skills.Text.V3.SentimentSkill) is the AI enrichment component that analyzes text and produces a sentiment score/label that can be written to the index. Option E is correct because a skill's output only reaches the target index field if an output field mapping in the indexer explicitly maps that enriched output (e.g., /document/sentiment) to the 'sentiment' field in the index. Option B is not required because normalizing sentiment values is optional and no custom skill is needed when the built-in skill already emits usable labels. Option C is not required because Azure AI Language sentiment analysis is already exposed through the built-in skill, so building a custom skill is unnecessary. Option D is incorrect because sentiment is a single scalar value (Edm.String or Edm.Double), not a Collection(Edm.String), so that field type would not properly store the skill output.

Answer

Implement a custom skill to normalize sentiment values.

Answer

Create a custom sentiment analysis skill using Azure AI Language.

Answer

Define a 'sentiment' field in the index with type Collection(Edm.String).

Question 6

You are using Azure AI Document Intelligence to extract data from scanned contracts. The contracts contain tables and handwritten signatures. Which TWO features should you enable?

Accepted Answer

Train a custom neural model to recognize handwritten signatures.. Option A is correct because a custom neural model in Azure AI Document Intelligence is specifically designed to handle unstructured and varied documents like scanned contracts, and it can be trained to recognize and extract handwritten signatures as a custom field, which is essential for this scenario. Option B is correct because enabling table extraction in the custom model allows the model to detect and extract tabular data from the contracts, which is a required capability since the contracts contain tables. Option C is not correct because OCR is automatically applied by Document Intelligence to read scanned text; it is not a separate feature that needs to be explicitly enabled for this scenario, and the question asks for features to enable. Option D is not correct because form recognition (key-value pair extraction) is a capability of prebuilt or custom models but does not specifically address the need for tables and handwritten signatures, and it is not a distinct feature to enable in this context. Option E is not correct because the prebuilt-layout model can extract tables and text but cannot recognize handwritten signatures as custom fields, and using it for all extraction would not meet the requirement for signature recognition.

Answer

Enable OCR to read scanned text.

Answer

Use form recognition to capture key-value pairs.

Answer

Use the prebuilt-layout model for all extraction.

Question 7

You are building a knowledge mining solution that uses Azure Cognitive Search and Azure AI Language. The solution must extract key phrases and detect the language of documents. Which THREE components are required?

Accepted Answer

A search index that contains fields for the extracted data.. Option B is correct because a search index is required to store and expose the enriched output, including fields for the extracted key phrases and detected language, so the data can be queried and returned. Option C is correct because the enrichment pipeline must be defined in a skillset, and the built-in Key Phrase Extraction skill and Language Detection skill from Azure AI Language (Cognitive Services) are the components that actually produce the key phrases and language values. Option D is correct because an Azure Cognitive Search data source is required to define the connection to the document store (for example, Azure Blob Storage) so the indexer knows where to pull the source documents from. Option A is not required because the built-in skills already produce the needed outputs, so a custom skill to combine them is unnecessary. Option E is not required because an indexer must exist to drive enrichment, but running it on a schedule is optional; an on-demand or one-time indexer execution is sufficient.

Answer

A custom skill to combine key phrases and language.

Answer

An indexer that runs on a schedule.

Question 8

You are designing a knowledge mining solution using Azure AI Search. The solution must process large volumes of PDFs daily. You need to minimize the cost of cognitive skills execution while ensuring the pipeline can handle transient failures. Which approach should you recommend?

Accepted Answer

Enable incremental enrichment on the indexer. Incremental enrichment allows the indexer to cache and reuse enrichments from previous runs, so only new or changed documents are processed by cognitive skills. This minimizes cost by avoiding redundant skill executions while also providing resilience against transient failures through the caching mechanism.

Answer

Disable field mappings

Answer

Increase the number of replicas

Answer

Use the free tier for the indexer

Question 9

Your team is building a custom ChatGPT-like copilot using Azure AI Foundry that answers questions based on internal HR policies stored in SharePoint. The solution must retrieve only the most relevant documents to minimize token usage. Which Azure AI Search feature should you configure?

Accepted Answer

Semantic ranking. Semantic ranking is the correct feature because it uses deep neural networks to re-rank search results based on semantic relevance to the query, ensuring only the most contextually appropriate documents are returned. This directly minimizes token usage by reducing the number of irrelevant documents passed to the copilot, which is critical for cost and performance in a ChatGPT-like system.

Answer

Synonyms

Answer

Scoring profiles

Answer

Filters

Question 10

A healthcare organization needs to mine clinical notes to find mentions of diseases, medications, and treatment procedures. The data is stored in Azure SQL Database. Which Azure AI service should they integrate with Azure AI Search to extract these entities?

Accepted Answer

Azure AI Health Insights. Azure AI Health Insights is specifically designed for healthcare scenarios and can extract medical entities such as diseases, medications, and treatment procedures from clinical notes. Integrating it with Azure AI Search allows for enriched search capabilities tailored to healthcare data. This service provides domain-specific natural language processing for clinical text.

Answer

Azure AI Document Intelligence

Answer

Azure AI Search

Answer

Azure AI Language

Question 11

Refer to the exhibit. You are configuring an Azure AI Search skillset. The skillset includes an EntityRecognitionSkill and a KeyPhraseExtractionSkill. After running the indexer, you notice that the 'organizations' field is empty in the index. What is the most likely cause?

Accepted Answer

The output field mapping for organizations is missing. In Azure AI Search skillsets, a skill's output must be explicitly mapped to an index field via an outputFieldMapping in the indexer definition. The EntityRecognitionSkill emits an 'organizations' output, but if no outputFieldMapping connects that enriched value to the 'organizations' field in the target index, the field remains empty even though the skill executed successfully. This is the most common cause of 'skill ran but field is blank' symptoms.

Answer

The skill output path is incorrect

Answer

The 'Organization' category is misspelled

Answer

The skills must be in reverse order

Question 12

Which TWO built-in cognitive skills in Azure AI Search can be used to extract entities from text?

Accepted Answer

PII Detection. Entity Recognition (E) is correct because it is a built-in cognitive skill in Azure AI Search that extracts entities such as people, organizations, locations, and quantities from text. PII Detection (B) is also correct because it is a built-in cognitive skill that extracts and identifies personal information entities like names, addresses, phone numbers, and email addresses from text. Custom Entity Lookup (A) is not correct because it is a custom skill that requires a user-defined list of terms, not a built-in cognitive skill for general entity extraction. Sentiment Analysis (C) is not correct because it evaluates positive/negative sentiment rather than extracting entities. Language Detection (D) is not correct because it identifies the language of the text, not entities within it.

Answer

Custom Entity Lookup

Answer

Sentiment Analysis

Answer

Language Detection

Question 13

You are a data engineer at a multinational corporation. The company has thousands of research reports in PDF format stored in Azure Blob Storage. The reports contain text, tables, charts, and handwritten annotations. Your team needs to build a knowledge mining solution using Azure AI Search that allows researchers to query the reports using natural language. The solution must extract text, table structures, and handwritten annotations. Additionally, the solution must handle multiple languages (English, Spanish, and French) and ensure that the index is updated daily as new reports are added. The search should prioritize the most recent reports. You have an Azure AI Search service in the S2 tier. Which combination of actions should you take to meet these requirements?

Accepted Answer

Use Azure AI Document Intelligence layout model with OCR, add a custom translation skill, and configure a scoring profile with freshness boosting. Using Azure AI Document Intelligence's layout and OCR capabilities extracts text, tables, and handwriting. The enrichment pipeline with a custom skill using the translation service handles multilingual content, and a scoring profile with freshness boosting prioritizes recent reports. Option A is incorrect because Azure AI Vision OCR alone does not extract table structure. Option C is incorrect because the Language service does not handle document layout. Option D is incorrect because scheduling the indexer once a week does not meet the daily update requirement.

Answer

Use Azure AI Vision OCR skill for text extraction, add a translation skill, and use a simple search query

Answer

Use Azure AI Document Intelligence prebuilt-read model, add a custom skill for language detection, and schedule the indexer weekly

Answer

Use Azure AI Language text extraction, a custom entity recognition skill, and enable semantic ranking

Question 14

You are developing a knowledge mining solution that extracts insights from customer feedback. Which TWO Azure AI services can be used to analyze the sentiment of the feedback and categorize it into topics?

Accepted Answer

Azure AI Language. Azure AI Language (option D) is correct because it provides prebuilt NLP capabilities such as sentiment analysis and key phrase extraction, and its topic modeling features can categorize customer feedback into topics. Azure AI Search with cognitive skills (option E) is correct because an AI enrichment pipeline can invoke cognitive skills like Sentiment and Key Phrase Extraction to analyze feedback and surface topics during indexing. Azure AI Personalizer (option A) is a reinforcement-learning service for ranking and recommending content, not for sentiment or topic analysis. Azure AI Translator (option B) only performs text translation and does not provide sentiment or topic categorization. Azure AI Custom Vision (option C) is an image classification/object detection service and cannot analyze text sentiment or topics.

Answer

Azure AI Personalizer

Answer

Azure AI Translator

Answer

Azure AI Custom Vision

Question 15

You are designing a knowledge mining solution for a large enterprise that uses Azure AI Search to index millions of documents. The solution must support high-availability and automatic failover. Which TWO actions should you take to meet these requirements?

Accepted Answer

Provision the Azure AI Search service in at least two regions.. Option C is correct because Azure AI Search high availability with automatic failover across regions requires deploying the service in at least two regions and using a multi-region architecture, since a single-region service cannot automatically fail over to another region. Option D is correct because within a region, provisioning at least two replicas ensures high availability and automatic failover of query and indexing workloads; Azure AI Search provides automatic failover when multiple replicas exist. Option A is incorrect because GRS applies to Azure Storage accounts, not to the Azure AI Search index data, and does not provide search service failover. Option B is incorrect because semantic search is a relevance feature (semantic ranking) and does not provide high availability or failover. Option E is incorrect because the text split skill is used during AI enrichment to chunk large documents for indexing and has no bearing on service availability or failover.

Answer

Use geo-redundant storage (GRS) for the index data.

Answer

Enable semantic search on the index.

Answer

Enable indexing of large documents using the text split skill.