AI-103 Text Analysis — Timed 15 Questions

Question 1

You are analyzing customer product reviews using the Azure AI Language REST API. You submit a batch of documents for sentiment analysis. One of the returned JSON objects includes a 'confidenceScores' property containing 'positive', 'neutral', and 'negative' values, but the overall document sentiment is listed as 'mixed'. What condition causes the service to return 'mixed'?

15 minute time limit — choose an answer to begin.

Accepted Answer

The confidence scores for both positive and negative sentiments exceed established thresholds within the same document.. The sentiment analysis service returns 'mixed' when the document contains both positive and negative sentiments with confidence scores that exceed the thresholds for each, indicating conflicting opinions within the same document. This typically happens when different sentences or aspects have opposing sentiments.

Answer

The document contains an equal number of uppercase and lowercase words.

Answer

The document length exceeds the 5,120 character limit.

Answer

The language detection confidence score is below 0.5.

Question 2

You are provisioning an Azure AI Language resource through the Azure CLI. You need to specify the resource kind parameter. Which value should you provide for a general text analysis resource?

Accepted Answer

TextAnalytics. When provisioning an Azure AI Language resource via the Azure CLI, the correct --kind value for a general text analysis resource is TextAnalytics. This kind maps to the Language service capabilities such as sentiment analysis, key phrase extraction, and NER. Specifying TextAnalytics ensures the resource exposes the expected Language APIs.

Answer

LanguageService

Answer

TextUnderstanding

Answer

CognitiveServices

Question 3

You are configuring abstractive summarization in Azure AI Language. You want to control the length of the generated summary output. Which parameter should you configure in your request body?

Accepted Answer

maxSummaryLength. In abstractive summarization for Azure AI Language, the maxSummaryLength parameter in the request body controls the maximum number of words (or tokens, depending on the API version) in the generated summary. Setting this parameter allows you to constrain the output length to meet specific requirements, such as a 100-word summary for an executive briefing. This is the only parameter among the options that directly controls summary length.

Answer

documentLengthLimit

Answer

sentenceCount

Answer

confidenceThreshold

Question 4

You are developing an application that processes multilingual documents using Azure AI Language. You want to ensure that your API payload complies with the maximum document size limit per single request document in synchronous calls. What is the maximum character length allowed for a single document input?

Accepted Answer

5,120 characters. Azure AI Language synchronous API calls accept a maximum of 5,120 characters per document in a single request. This limit applies to synchronous document analysis operations and must be respected to avoid request failures.

Answer

1,000,000 characters

Answer

50,000 characters

Answer

1,024 characters

Question 5

You are integrating Azure AI Language into a secure enterprise application. Management mandates that all data sent to the API must remain within your Azure region and that no customer data is used to train Microsoft's foundational models. Which feature of Azure AI Language supports this?

Accepted Answer

Using custom text classification with customer-managed keys (CMK) and disabling public network access.. Custom text classification with customer-managed keys (CMK) and disabled public network access ensures data is encrypted with keys you control and only reachable via private endpoints within your Azure region, satisfying data-residency and no-training requirements. Azure AI Language's 'no training on customer data' guarantee plus CMK and Private Link gives the compliance posture management demands. This combination directly addresses both regional containment and data sovereignty.

Answer

Switching from synchronous to asynchronous batch processing.

Answer

Using the Free (F0) pricing tier.

Answer

Enabling abstractive summarization with cross-region replication.

Question 6

You are analyzing legal contracts using Azure AI Language custom named entity recognition. You notice that overlapping entities (e.g., 'Acme Corporation' inside a larger entity 'Acme Corporation Headquarters') are causing extraction errors in your custom model. How should you resolve this in Language Studio?

Accepted Answer

Configure the project schema to support overlapping entities during project creation.. When configuring custom NER projects, you can enable or configure support for overlapping entities during project creation if your schema requires nested or overlapping extractions.

Answer

Increase the confidence score threshold to 0.99.

Answer

Switch the project type from custom NER to custom text classification.

Answer

Convert all contract texts to lowercase before tagging.

Question 7

You need to detect the primary language of customer feedback submissions before routing them to downstream translation and analysis pipelines. Which Azure AI Language API endpoint should your application call?

Accepted Answer

/languages/detect. The Azure AI Language language detection feature is exposed at the '/languages/detect' endpoint (part of the text analytics API), which returns the detected language name and ISO 639-1 code plus a confidence score. This is the correct endpoint for identifying the primary language of customer feedback before routing to translation or analysis.

Answer

/keyphrases

Answer

/language/:analyze-text

Answer

/sentiment

Question 8

You are using the Azure AI Language REST API to perform extractive summarization on a large collection of research papers. You want to prioritize the top sentences that represent the core findings. Which parameter can you adjust in the API request to control the maximum number of sentences returned in the summary?

Accepted Answer

maxSentenceCount. In the Azure AI Language extractive summarization API, the 'maxSentenceCount' parameter in the request body controls the maximum number of sentences returned in the summary. Setting it limits the output to the top-ranked sentences that best represent the document's core content, which is exactly what the scenario requires.

Answer

summaryLengthLimit

Answer

topKeyPhrases

Answer

minConfidenceScore

Question 9

You are developing a .NET application that analyzes customer feedback. Which NuGet package should you install to use the Azure AI Language client library in C#?

Accepted Answer

Azure.AI.TextAnalytics. The current Azure AI Language (Text Analytics) client library for .NET ships as the NuGet package Azure.AI.TextAnalytics, part of the Azure SDK for .NET. Installing it gives you the TextAnalyticsClient class used to call sentiment analysis, key phrase extraction, entity recognition, and other Language service features. The package name follows the Azure SDK convention of Azure.AI.<Service>.

Answer

Azure.AI.Language.Text

Answer

Microsoft.CognitiveServices.TextAnalytics

Answer

Microsoft.Azure.CognitiveServices.Language

Question 10

You are building an application that extracts named entities from technical specifications. You need to ensure that entity extraction respects casing requirements for part numbers (e.g., 'Model-X' vs 'Model-x'). Where can you configure case sensitivity for custom NER projects?

Accepted Answer

In the project settings within Language Studio during project creation.. Case sensitivity for custom Named Entity Recognition (NER) projects in Azure AI Language is configured as a project-level setting within Language Studio during project creation. This setting determines whether the model treats tokens like 'Model-X' and 'Model-x' as distinct entities during training and inference. Because it is a project-wide configuration, it applies consistently across all API calls made to that project's deployment endpoint, rather than being specified per request.

Answer

In the HTTP request headers of every API call.

Answer

In the Azure Monitor diagnostics logs.

Answer

In the Azure portal subscription access control blade.

Question 11

You need to detect the language of a batch of text documents and want to minimize the number of API calls. What is the maximum number of documents you can include in a single request payload to the language detection endpoint?

Accepted Answer

10 documents. The Azure AI Language language detection endpoint accepts a maximum of 10 documents per request payload. To minimize API calls while respecting this limit, you batch documents in groups of up to 10. This is a documented service limit that candidates must know for exam scenarios involving batching.

Answer

1 document

Answer

1,000 documents

Answer

100,000 documents

Question 12

You are building an application to process customer reviews. Which TWO metrics are returned in the standard JSON response of a sentiment analysis document evaluation? (Choose two.)

Accepted Answer

Overall document sentiment label. Option A is correct because a sentiment analysis document evaluation returns an overall document sentiment label (for example, POSITIVE, NEGATIVE, NEUTRAL, or MIXED) that summarizes the sentiment of the entire document. Option E is correct because the standard JSON response also includes sentiment confidence scores for the positive, neutral, and negative categories, giving the probability or confidence level for each sentiment class. Options B, C, and D are not part of the standard sentiment analysis response: estimated reading time, detected language translation confidence, and total word count are unrelated metrics that would come from other services or features, not from a sentiment analysis document evaluation.

Answer

Estimated reading time in minutes

Answer

Detected language translation confidence score

Answer

Total word count of the document

Question 13

You are troubleshooting an Azure AI Language resource deployment. Which THREE error codes or HTTP status codes might indicate an issue with your API request structure or payload? (Choose three.)

Accepted Answer

404 Not Found. Option A (404 Not Found) is correct because it indicates the request was sent to an invalid or nonexistent endpoint path, which is a structural problem with how the API request URL was constructed for the Azure AI Language resource. Option C (400 Bad Request) is correct because this status code is returned when the request is malformed, such as invalid JSON syntax, missing required headers like Content-Type or Ocp-Apim-Subscription-Key, or improperly formatted parameters. Option E (422 Unprocessable Entity) is correct because it signals that the request was syntactically valid but semantically incorrect, such as a payload containing fields with invalid values or unsupported data types for the Azure AI Language API. Option B (100 Continue) is not correct because it is an informational interim response indicating the client may continue sending the request body, not an error related to request structure. Option D (200 OK) is not correct because it indicates a successful request, meaning no structural or payload issue occurred.

Answer

100 Continue

Answer

200 OK

Question 14

You are troubleshooting a custom text classification project in Language Studio where your model's macro F1-score is low. Which THREE remedial actions can help improve the macro F1-score across all categories? (Choose two.)

Accepted Answer

Review and correct inconsistent or ambiguous labels in the training dataset.. Improving macro F1-score requires addressing imbalances, adding more training data for poorly performing classes, and reviewing labeling consistency. (Note: Since this is a 'Which THREE' question, select three valid remediation options).

Answer

Manually hardcode all possible classification outputs in client application code.

Answer

Delete all training data across every category.

Question 15

You are building an Azure AI Language pipeline that processes supplier contracts written in English and Spanish. Before you can analyze the text, you must detect which language each document is written in so you can route it to the correct downstream analyzer. Which TWO actions must your solution perform when calling the language detection capability? (Choose two.)

Accepted Answer

Provide a 'countryHint' value only when the text is too short or ambiguous for reliable detection.. Language detection requires submitting the document text and interpreting the returned language name and confidence score, which drives the routing decision. Because short or ambiguous passages can be misclassified, supplying an optional country hint only in those cases improves accuracy. Preview model pinning, single-document batching, and the logging opt-out flag do not affect detection quality or routing.

Answer

Set the 'model-version' parameter to the latest preview version to ensure Spanish is supported.

Answer

Set the 'kind' property to 'LanguageDetection' and include the 'loggingOptOut' flag to suppress telemetry.

Answer

Send the documents in batches of exactly one document per request to guarantee language detection accuracy.